Home Blog Page 653

97 Million Downloads and Growing Crypto Infrastructure From Bitgo to Coingecko – Crypto News Bitcoin News

info@uweb3.io
-
0
97 Million Downloads and Growing Crypto Infrastructure From Bitgo to Coingecko – Crypto News Bitcoin News

AI Tool Integration Just Got Standardized — MCP Downloads Cross 97 Million in March 2026

Anthropic open-sourced MCP on Nov. 25, 2024, alongside reference servers for Google Drive, Slack, Github, and Postgres, with native support built into Claude Desktop. Early adopters included Block and Apollo; IDEs such as Zed, Replit, Codeium, and Sourcegraph began integration within weeks.

The protocol defines how AI models connect to external tools, databases, APIs, and workflows. An MCP host, such as Claude Desktop, ChatGPT, or VS Code Copilot, communicates with an MCP server, a lightweight wrapper around a specific tool or data source. One server can serve every compatible client without per-model custom code.

The official project site describes it as “a USB-C port for AI applications.” That framing captures the practical shift: instead of building separate connectors for each AI platform, developers expose a single MCP server and gain compatibility across Claude, ChatGPT, Gemini, Microsoft Copilot, and any other MCP-compatible client.

OpenAI added full MCP support across ChatGPT and its Agents SDK in March 2025, which analysts have identified as an inflection point in adoption. Google, Microsoft, AWS, and dozens of other platforms followed through mid-2025.

In December 2025, Anthropic donated MCP to the newly formed Agentic AI Foundation (AAIF) under the Linux Foundation. OpenAI and Block joined as co-founders. Platinum members include AWS, Google, Microsoft, Cloudflare, Github, and Bloomberg. The governance structure mirrors that of Kubernetes and Pytorch — vendor-neutral and community-managed.

As of March 2026, more than 10,000 active MCP servers exist across public and enterprise deployments. The combined Python and Typescript SDKs account for the 97 million monthly download figure, up from roughly 100,000 at launch in late 2024. The crypto sector has moved quickly to build MCP infrastructure. Bitgo launched an official MCP server in March 2026, enabling AI tools and development environments to interact with its institutional digital asset custody platform via natural language.

Coinbase released a Payments MCP through its Developer Platform in late 2025, connecting AI agents to crypto wallets, onramps, and stablecoin transactions. Crypto.com released a Market Data MCP server providing live price quotes, order books, and candlestick charts. Coingecko launched its own server supplying real-time data for more than 15,000 cryptocurrencies and 1,000-plus exchanges.

Cross-chain protocol Debridge deployed an MCP server in February 2026, enabling non-custodial swaps and bridging across EVM chains and Solana. Enterprises building internal AI tooling have shifted away from one-off API connectors. Publishing a single MCP server gives a product instant interoperability with every major AI client, a network effect that proprietary integration approaches cannot replicate.

Security researchers have flagged that many public MCP servers have not undergone formal audits. The Linux Foundation stewardship brings standardized authentication and transport requirements, but deployment-level security remains the responsibility of individual server maintainers.

The protocol reached that comparison point in under 18 months. The MCP specification, SDKs, and reference implementations are available at github.com/modelcontextprotocol. The core documentation lives at modelcontextprotocol.io.

FAQ 🤖

  • What is the Model Context Protocol (MCP)? MCP is an open-source protocol that standardizes how AI applications connect to external tools, databases, and APIs without requiring custom integrations for each model.
  • How many monthly downloads does the MCP SDK have in 2026? The combined Python and TypeScript MCP SDKs reached approximately 97 million monthly downloads as of March 2026.
  • Which AI platforms support MCP? Claude, ChatGPT, Gemini, Microsoft Copilot, Cursor, and VS Code Copilot all ship native MCP support as of early 2026.
  • Which crypto companies have launched MCP servers? BitGo, Coinbase, Crypto.com, CoinGecko, and deBridge have each released official MCP servers connecting AI agents to crypto data and transaction infrastructure.

NYSE Owner ICE Pours Another $600 Million Into Polymarket

info@uweb3.io
-
0
NYSE Owner ICE Pours Another $600 Million Into Polymarket

Intercontinental Exchange has now deployed nearly $2 billion into the onchain prediction market, underscoring Wall Street’s growing conviction that event-based trading is here to stay.

Intercontinental Exchange, the parent company of the New York Stock Exchange, on Friday announced a new $600 million direct cash investment in Polymarket, completing the exchange operator’s structured investment arrangement with the prediction market platform.

The investment is part of a broader equity capital fundraise by Polymarket, according to a press release from ICE. The company also expects to purchase up to $40 million in Polymarket securities from existing holders, which would close out its obligations under the deal first announced in October 2025. The valuation of Friday’s investment is expected to be disclosed after Polymarket completes its fundraising.

ICE made an initial $1 billion direct investment in Polymarket at that time, in what was the largest single investment ever made in a prediction market company. That deal valued Polymarket at roughly $8 billion pre-investment and established ICE as a global distributor of Polymarket’s event-driven data.

ICE’s interest in Polymarket extends beyond a passive equity stake. In February, ICE launched the Polymarket Signals and Sentiment Tool, a product that normalizes real-time and historical prediction market data into structured feeds for institutional traders. The tool packages Polymarket’s crowd-sourced probability assessments as market signals alongside traditional financial instruments.

Prediction Market Arms Race

The capital injection comes amid an unprecedented wave of institutional investment into prediction markets. Rival platform Kalshi raised approximately $1 billion at a $22 billion valuation earlier this month in a round led by Coatue Management. Polymarket is reportedly targeting a valuation of around $20 billion in its current round, according to The Wall Street Journal.

Prediction market monthly volumes have grown 130-fold since early 2024, making it one of the fastest-growing categories in finance. Open interest across platforms crossed $1 billion for the first time in February.

Regulatory Crosswinds

The investment arrives against a complex regulatory backdrop. The CFTC recently issued an advance notice of proposed rulemaking signaling its intent to build a comprehensive regulatory framework for prediction markets. Meanwhile, some lawmakers have introduced legislation that would block prediction markets from offering contracts on war and sports outcomes.

At the state level, regulators continue to challenge the industry — Arizona’s attorney general recently filed criminal charges against Kalshi, alleging it operates an illegal gambling business in the state.

Still, institutional capital appears undeterred by the regulatory uncertainty. For ICE, the completion of its nearly $2 billion investment arrangement signals that one of the world’s largest market infrastructure operators views prediction markets not as a passing novelty but as a category that may eventually sit alongside equities, futures, and fixed income.

This article was written with the assistance of AI workflows. All our stories are curated, edited and fact-checked by a human.

Gems Trade integrates Fireblocks MPC security to strengthen custody controls and eliminate single-key risk

info@uweb3.io
-
0
Gems Trade integrates Fireblocks MPC security to strengthen custody controls and eliminate single-key risk

Cyprus, March 24, 2026—Gems Trade, a next-generation centralized cryptocurrency exchange (CEX) and the core trading hub of the Gems Ecosystem, announces its partnership with Fireblocks, the world’s most trusted digital asset infrastructure company. Through this partnership, Fireblocks will provide the custody and transaction infrastructure to secure digital assets and control how they move across the exchange, giving Gems Trade stronger oversight and operational control.

As the crypto exchange market matures, competition is no longer driven solely by listings. Users, market makers, and institutions increasingly evaluate platforms based on the strength of the infrastructure supporting them and whether it can mitigate risk before a single incident escalates into a platform-wide disruption. However, many exchanges still rely on fragmented security setups, where wallet management, approvals, and monitoring operate in silos. These disconnected environments can slow response times, increase operational risk, and weaken user confidence, especially during periods of peak volatility.

To address these challenges, Gems Trade is integrating Fireblocks’ multi-party computation (MPC) security architecture, which reduces single-key risk by distributing transaction authorization across multiple secure systems. Operationally, this means transfers follow defined approval steps, while monitoring tools track activity and maintain auditable transaction records. 

This approach helps protect against internal and external threats and supports operational continuity during technical disruptions or unexpected events. As the platform scales, the infrastructure is designed to support the security standards of retail and professional traders, market makers, and institutions.

This partnership reflects Gems Trade’s broader strategy to build a platform where performance, security, and trust converge. Evolving from the broader Gems ecosystem, Gems Trade has expanded into a full trading venue powered by the $GEMS token, bringing traders, projects, and investors together in a single, high-performance environment.

“As an exchange, we know that we are only as strong as the infrastructure behind us,” said Omri Hanover, Head of Project at Gems Trade. “We’re proud to partner with one of the most trusted infrastructure providers in crypto, strengthening the foundation behind our exchange as we enter our next phase of growth. Security will always be our top priority, and this collaboration supports a safer, more resilient environment for everyone engaging with the Gems ecosystem.”

“Gems Trade is building for a market that increasingly demands both performance and security at the infrastructure level,” said Omer Amsel, Head of Web3 at Fireblocks. “We’re proud to support that growth by providing the security and operational layer the team can build on as they scale.”

Morgan Stanley enters bitcoin ETF race with market-leading low fee

info@uweb3.io
-
0
Morgan Stanley enters bitcoin ETF race with market-leading low fee

Morgan Stanley plans to price its proposed spot bitcoin BTC$65,998.68 exchange-traded fund (ETF) at 14 basis points, a level just below current low-cost options for similar products, according to an amended filing with the U.S. Securities and Exchange Commission (SEC). The move could set off a new round of fee competition among existing funds.

The latest S-1 filing, filed Friday, shows the bank undercutting rivals that charge closer to 15 to 25 basis points. The lowest fee on the market today is Grayscale’s Bitcoin Mini Trust ETF BTC$65,998.68, which carries a 0.15% expense ratio. Larger funds, including BlackRock’s iShares Bitcoin Trust (IBIT), priced their products at 25 basis points.

On paper, the gap looks narrow. In practice, it may be enough to shift money.

Spot bitcoin ETFs offer near-identical exposure. Each fund holds bitcoin and aims to track its price. That leaves cost as one of the few variables investors and advisors can act on. A financial advisor can move a client from one ETF to another with a single trade, keeping the same exposure while lowering annual fees.

That dynamic has shaped the ETF market before, and lower-cost products tend to attract inflows, while higher-fee funds can see assets drift out over time. Grayscale’s flagship product, its Bitcoin Trust (GBTC), holds about $10 billion in assets, down from $29 billion at launch in January 2024.

Morgan Stanley’s scale adds another layer. Its wealth management arm oversees trillions in client assets and has one of the largest adviser networks in the industry. Even small allocation changes across that base could move billions of dollars between funds.

The pricing decision also points to strategy. By entering with a lower fee, Morgan Stanley may be aiming to quickly gain share in a market where products are hard to differentiate. Cost and access, not structure, often decide which funds grow.

The filing follows confirmation from the New York Stock Exchange that it has issued a listing notice for MSBT, signaling the product could begin trading quickly if approved.

If regulators sign off, the fund would be the first spot bitcoin ETF issued directly by a major U.S. bank, setting up a new phase of competition where fees and distribution drive the outcome.

Survey Shows Institutions Want Solana Over XRP And Dogecoin, Here Are The Figures

info@uweb3.io
-
0
Survey Shows Institutions Want Solana Over XRP And Dogecoin, Here Are The Figures

Trusted Editorial content, reviewed by leading industry experts and seasoned editors. Ad Disclosure

A recent survey by Coinbase and EY-Parthenon shows that institutional investors are more allocated to Solana over XRP and Dogecoin. This contrasts with the current trend in spot crypto ETFs, where XRP ETFs boast more net assets than SOL and DOGE ETFs. 

Institutions Are More Invested In Solana Than XRP And Dogecoin

The survey shows that more institutions are investing in Solana than XRP and Dogecoin. 36% of these participants had allocations to SOL as of January 2026, while 38% plan to add to their allocations. Meanwhile, 18% allocated to XRP as of January, while 25% plans to add the token to their allocations this year. 

Dogecoin is far behind Solana and XRP, with 2% of these institutions investing in DOGE as of January 2026, while 2% plan to add the meme coin to their allocations. It is worth noting that SOL is only behind Bitcoin and Ethereum and is well ahead of Chainlink, Binance Coin, Cardano, Tron, and Bitcoin Cash. 

Solana
Source: Chart from Cftassets

This survey contrasts with the current trend among crypto ETFs, showing that investors allocate more to XRP ETFs than to Solana and Dogecoin ETFs. SoSoValue data shows that the XRP ETFs currently boast net assets of $949.15 million, representing 1.14% of the XRP’s market cap. Meanwhile, the Solana and Dogecoin ETFs boast net assets of $849.65 million and $9.12 million, respectively. 

Furthermore, the XRP ETFs have seen more inflows since they launched than the Solana and Dogecoin ETFs. The XRP ETFs currently boast total net inflows of $1.21 billion, while the SOL and DOGE ETFs have seen inflows of $993.38 million and $7.64 million, respectively. 

Institutions Holding Spot ETFs Over Spot Crypto 

The survey also showed that most of these institutions are gaining crypto exposure through the crypto ETFs rather than holding spot crypto. As of January 2025, 64% of these institutions held spot crypto ETFs to gain exposure to Solana, XRP, Dogecoin, and other digital assets. This figure has climbed to 66% as of January 2026, signaling that more institutions are investing in crypto amid regulatory clarity. 

Furthermore, 39% of these institutions held spot crypto as of January 2025. However, this figure has decreased to 36% as of January 2026, suggesting that institutions prefer to gain crypto exposure through an ETF wrapper rather than holding crypto directly. These institutions have also been seeking crypto exposure through the digital asset treasury companies (DATs). As of January 2025, 51% of these institutions invested in these DATs, and that figure increased to 53% as of January 2026. 

At the time of writing, the XRP price is trading at around $1.36, down over 2% in the last 24 hours, according to data from CoinMarketCap.

Solana
SOL trading at $85 on the 1D chart | Source: SOLUSDT on Tradingview.com

Featured image from Freepik, chart from Tradingview.com

Editorial Process for bitcoinist is centered on delivering thoroughly researched, accurate, and unbiased content. We uphold strict sourcing standards, and each page undergoes diligent review by our team of top technology experts and seasoned editors. This process ensures the integrity, relevance, and value of our content for our readers.

NYSE Parent Company Finalizes Polymarket Investment, Totaling $1.6 Billion

info@uweb3.io
-
0
NYSE Parent Company Finalizes Polymarket Investment, Totaling $1.6 Billion

In brief

  • ICE has invested another $600 million into Polymarket, fulfilling its commitment made in October.
  • Rival Kalshi recently raised $1 billion at a $22 billion valuation, outpacing Polymarket’s current valuation.
  • Prediction markets face mounting regulatory pressure, with lawmakers moving to ban insider trading on the platforms.

New York Stock Exchange parent company Intercontinental Exchange has completed its investment into prominent prediction market platform Polymarket, with the final total landing at $1.6 billion.

ICE said the new funding is part of an equity capital fundraising by Polymarket, and that the firm intends to purchase up to $40 million worth of Polymarket securities from existing holders.

The NYSE parent company made a commitment of up to $2 billion to Polymarket in October 2025 that valued the company at $9 billion. Back then, the company made a $1 billion initial investment. The additional $600 million and the plan to purchase securities from existing investors mean that the firm’s obligations to Polymarket have now been fulfilled.

Polymarket has been locked in a heated competition with rival platform Kalshi, even when it comes to fundraising.

Kalshi just raised $1 billion earlier this month in a round led by Coatue Management, at a $22 billion valuation—double its $11 billion valuation from a December round backed by Paradigm, Andreessen Horowitz, Ark Invest, and Sequoia.

Kalshi has been on a rapid fundraising tear since winning a CFTC court battle in May 2025. That cleared the way for its election contracts to be offered and the company to scale from a $2 billion valuation in June 2025 to its current $22 billion in under a year.

Polymarket recently put together a 3-day Washington D.C. pop-up experience, the Situation Room, which was billed as the world’s first brick-and-mortar destination for monitoring global prediction markets. It got mixed reviews from journalists in attendance—tech outlet Wired called it “a disaster,” due to the screens being off on opening night thanks to technical difficulties.

The investment comes as prediction markets face growing regulatory scrutiny in Washington and in multiple states.

Massachusetts Rep. Seth Moulton banned his staff from trading on platforms like Polymarket and Kalshi this week, citing concerns about insider trading. The additional funding for Polymarket arrives a few weeks after bipartisan lawmakers introduced the PREDICT Act to extend similar restrictions to members of Congress, senior officials, and their families.

Separately, senators have proposed bans on sports contracts and war-related markets, following controversy over profitable bets tied to U.S. strikes on Iran and the capture of Venezuela’s Nicolás Maduro. Also on Friday, California Governor Gavin Newsom signed an executive order to ban state officials and governor appointees from betting on prediction markets using insider info.

Daily Debrief Newsletter

Start every day with the top news stories right now, plus original features, a podcast, videos and more.

ECB Study Concludes DeFi DAOs Aren’t as Decentralized as They Claim

info@uweb3.io
-
0
ECB Study Concludes DeFi DAOs Aren’t as Decentralized as They Claim

A new working paper from the European Central Bank examined four major protocols and found that a small number of actors control the bulk of governance token holdings.

A European Central Bank working paper challenges the notion that decentralized autonomous organizations (DAOs) deliver on their promise of distributed governance, finding that token holdings and voting power across four major DeFi protocols are heavily concentrated among a handful of actors.

The study examined governance structures at Aave, MakerDAO, Ampleforth, and Uniswap using data from late 2022 and mid-2023. The researchers analyzed the top 100 token holders and top 20 voters for each protocol, reviewed 248 governance proposals, and attempted to trace the real-world identities behind pseudonymous blockchain addresses.

The findings land at a moment when governance disputes are roiling some of the very protocols examined in the study, and DeFi projects more broadly are grappling with whether the Labs-plus-DAO structure is fit for purpose.

Top 100 Holders Command Over 80% of Supply

Across all four protocols, the top 100 holders controlled more than 80% of the total governance token supply during both snapshot periods. At Aave and Uniswap, the top five accounted for roughly half of all holdings. MakerDAO was the relative outlier, with the top five holding around 36%.

The concentration proved sticky over time, with distributions remaining largely unchanged between October 2022 and May 2023.

When the researchers dug into who sits behind the top addresses, they found that for most protocols, roughly half or more of holdings traced to addresses associated with the protocols themselves — encompassing treasuries, founders, and developer allocations — or to centralized and decentralized exchanges.

Protocol-associated addresses held 43% of Uniswap’s UNI supply. Centralized exchange holdings were particularly notable at Aave (16%) and Ampleforth (19%). Binance emerged as the dominant exchange holder across all four protocols, with holdings ranging from 2% to 15% of total supply.

The researchers cautioned that available data doesn’t distinguish between tokens held by exchanges on their own behalf versus those held in custody for customers.

Delegates Dominate Voting

The most active voters on governance proposals turned out to be predominantly delegates — entities to whom smaller token holders assign their voting power. This dynamic has long been a known issue in DAO governance, where low voter turnout and outsized whale participation leave a small group of recurring participants shaping protocol decisions.

The top voter at Uniswap in both snapshots was a16z, the venture capital firm, which saw its delegator count grow from 100 to 125 over the study period. At Aave, the protocol’s own smart contracts held the top-voter position.

Of the 68 top voters identified across all protocols, the researchers could not determine the identities of roughly one-third to nearly half of them. Among those they could identify, individuals made up about 21%, followed by Web3 companies at 19%, university blockchain societies, and VC firms.

Uniswap had the highest delegation rate at 27%, with its top 18 voters holding more than half the delegated power.

The ECB team also systematically categorized the 248 proposals and found that “risk parameters” — covering loan-to-value ratios, liquidation thresholds, borrowing rates, and debt ceilings — were the most common, accounting for 28%. Asset listing proposals made up 23%.

Implications for Regulation

The findings carry direct implications for the ongoing policy debate over how to regulate DeFi. The EU’s Markets in Crypto-Assets regulation exempts services provided in a “fully decentralized manner,” but the ECB researchers argue the protocols they studied fall well short of that standard.

Governance token holders, protocol developers, and centralized exchanges have frequently been proposed as potential regulatory entry points. However, the researchers concluded that the ambiguity surrounding who actually controls governance makes all three difficult to use in practice.

“It is not always clear who in the end is responsible or can be held accountable based on publicly available data,” the authors wrote.

The paper also drew parallels between DeFi governance and traditional corporate shareholder governance, noting that both systems suffer from low voter turnout and outsized influence by a small number of recurring participants.

But DeFi lacks the institutional safeguards — proxy voting rules, stewardship codes, disclosure requirements, and fiduciary obligations — that help mitigate those dynamics in public companies. As DAOs increasingly adopt formal legal structures, the researchers suggested that hybrid models integrating traditional legal frameworks with blockchain-based governance may ultimately be needed.

This article was written with the assistance of AI workflows. All our stories are curated, edited and fact-checked by a human.

How AI is Rewiring Global B2B Commerce

info@uweb3.io
-
0
How AI is Rewiring Global B2B Commerce

The financial technology landscape is undergoing a profound paradigm shift. For the past two years, the conversation has been dominated by Generative AI—models designed to draft emails, write code, and synthesize data. However, the true disruption for global commerce lies in the next evolutionary step: Agentic AI. We are moving from artificial intelligence that simply “advises” to AI that “executes.”

In the financial sector, this transition gives rise to Agentic Payments—a framework where autonomous AI agents negotiate pricing, select optimal routing networks, and execute B2B settlements with zero human intervention. As enterprises increasingly deploy AI to manage supply chains and procurement, traditional payment gateways are proving inadequate. The market now demands a fundamentally new payment infrastructure—one built natively for machine-to-machine (M2M) interaction.

The Friction in Traditional Global Payments

To understand the necessity of agentic payments, we must examine the bottlenecks of the current B2B payment architecture. Today’s global commerce is still hindered by fragmented banking networks, batch-processing delays, and UI-heavy portals designed for human operators.

When an enterprise deploys an AI agent to optimize its inventory, the AI can predict stock shortages and independently order materials from a supplier. Yet, when it comes to the final step—moving the funds—the autonomous workflow breaks. The AI hits a wall of manual approvals, complex KYC/AML check-boxes, and incompatible banking APIs. Traditional payment rails were built for human pacing, requiring manual data entry, physical security tokens, or complex multi-step authentications that an AI agent simply cannot navigate efficiently.

For autonomous commerce to scale, the payment layer must become invisible, instant, and entirely API-driven. AI agents do not need user-friendly dashboards; they need robust, machine-readable financial protocols that allow them to query balances, execute transactions, and reconcile ledgers in milliseconds.

Building the Foundation: Trust, Security, and Programmability

The most significant hurdle in adopting agentic payments is not technological, but psychological and regulatory: How do we trust an AI with the corporate treasury?

The answer lies in Programmable Payments and strict Autonomy Gates. Before an AI agent can execute a transaction, the underlying financial infrastructure must support highly granular, programmable logic. Enterprise finance teams need the ability to hard-code spending limits, velocity constraints, and approved counterparties directly into the payment rail.

For instance, an AI agent might be granted the autonomy to pay cloud infrastructure bills up to $50,000 automatically, but any payment exceeding that threshold, or directed to a newly onboarded vendor, would trigger a smart contract requiring cryptographic human-in-the-loop (HITL) approval.

Furthermore, compliance must be shifted left. Modern payment networks must integrate real-time, AI-driven KYC and AML screening directly into their APIs, ensuring that every micro-transaction executed by an autonomous agent is instantly audited and fully compliant with international financial regulations.

Pioneering the Infrastructure for AI Commerce

Recognizing this seismic shift, the most forward-thinking fintech platforms are rapidly re-architecting their systems. The goal is no longer just moving money, but providing the orchestration layer for autonomous financial operations (FinOps).

A prime example of this evolution is PhotonPay, a global payment platform that is actively pivoting its infrastructure to support agentic workflows. Recognizing that the future of B2B commerce will be driven by software agents, PhotonPay is developing deeply programmable, API-first payment rails designed specifically for machine execution.

Rather than relying on legacy batch processing, platforms evolving in this direction focus on real-time data synchronization and smart routing. PhotonPay’s architecture is being engineered to allow enterprise AI systems to seamlessly plug into a unified global treasury. This means an AI procurement agent could theoretically use the platform’s API to analyze real-time liquidity, split a massive vendor payment into multiple optimized tranches, and execute the settlement instantly—all while adhering to pre-defined corporate governance rules.

By building native interoperability for AI agents, PhotonPay is addressing the critical missing link in autonomous commerce: an intelligent, secure, and fully programmable financial execution layer.

The Road Ahead: Embracing Autonomous FinOps

The transition to agentic payments is not a distant futuristic concept; the foundational building blocks are being deployed today. As AI agents become standard components of enterprise ERP systems, supply chain management, and corporate treasuries, the friction of legacy payment rails will become an unacceptable business liability.

For fintech leaders, banking executives, and corporate CFOs, the mandate is clear. The next decade of financial technology will not be won by those who build the best user interfaces for humans, but by those who build the most secure, programmable, and scalable payment infrastructure for machines.

The dawn of agentic commerce is here. It is time to ensure your payment stack is ready for the autonomous future.

Anthropic’s ‘Most Capable’ AI Model Claude Mythos Leaks, Deemed Major Cybersecurity Threat

info@uweb3.io
-
0
Anthropic’s ‘Most Capable’ AI Model Claude Mythos Leaks, Deemed Major Cybersecurity Threat

In brief

  • A leaked draft post revealed Anthropic’s most powerful AI model, Claude Mythos.
  • The model also appears to introduce a new tier above Opus, internally referred to as “Capybara.”
  • Cybersecurity stocks declined after reports suggested the system could accelerate AI-driven cyberattacks.

Claude creator Anthropic is developing a new AI model called Claude Mythos, described internally as the company’s most capable model to date, with draft materials about the system being leaked online this week.

The existence of the model was first reported by Fortune on Thursday after unpublished files tied to Anthropic’s blog were discovered in a publicly accessible data cache. An Anthropic spokesperson confirmed the existence of the model to the publication.

“We’re developing a general purpose model with meaningful advances in reasoning, coding, and cybersecurity,” an Anthropic spokesperson told Fortune. “Given the strength of its capabilities, we’re being deliberate about how we release it. As is standard practice across the industry, we’re working with a small group of early access customers to test the model. We consider this model a step change and the most capable we’ve built to date.”

In an archived development page reviewed by Decrypt, Anthropic called Mythos “the most powerful AI model we’ve ever developed.”

“Mythos is a new name for a new tier of model: larger and more intelligent than our Opus models—which were, until now, our most powerful,” Anthropic wrote. “We chose the name to evoke the deep connective tissues that link together knowledge and ideas.”

According to Anthropic, Mythos scored “dramatically higher” than Claude Opus 4.6 on tests of software coding, academic reasoning, and cybersecurity.

The leak of Mythos appears to have originated from draft materials stored in an unsecured content management system. According to Fortune, Anthropic restricted public access to the data store after being notified that the files were searchable online. The company attributed the exposure to human error in the configuration of its CMS tools.

However, Anthropic’s documents labeled Mythos as version one of the new model, and described version two internally as “Capybara,” which the company also positioned above its current top-tier Opus models.

The draft materials also highlighted concerns about the system’s potential cybersecurity implications.

“Although Mythos is currently far ahead of any other AI model in cyber capabilities, it presages an upcoming wave of models that can exploit vulnerabilities in ways that far outpace the efforts of defenders,” the company wrote.

Because of those risks, the company said it plans to release the model cautiously, beginning with a limited early-access rollout aimed at organizations working on cybersecurity defense.

Anthropic did not immediately respond to Decrypt’s request for comment.

While Anthropic took down the blog post, news of the leak quickly spilled into financial markets.

Shares of several cybersecurity firms dropped after the reports surfaced, including Palo Alto Networks (PANW), which fell about 7%, and CrowdStrike (CRWD), which dropped roughly 6.4%. Meanwhile, Zscaler (ZS) declined around 5.8%, and Fortinet (FTNT) slipped about 4% during Friday trading, according to Yahoo Finance.

The selloff reaction echoes a similar market response to the reveal of a new Anthropic product. In February, Anthropic unveiled Claude Cowork, an AI system designed to automate complex workplace tasks—including contract review and compliance—which triggered a broad sell-off across software and professional-services companies.

That sell-off erased roughly $285 billion in market value as investors reassessed the long-term impact of AI agents on enterprise software businesses.

“The market’s response was a signal, not that AI agents will immediately replace these businesses, but that investors are finally pricing in the structural risk that foundation model providers can now compete directly with the software layer,” Nexatech Ventures founder Scott Dylan told Decrypt at the time. “That’s a polite way of saying if Anthropic can build a legal workflow tool in-house, what’s stopping them from doing the same for finance, procurement, or HR?”

Daily Debrief Newsletter

Start every day with the top news stories right now, plus original features, a podcast, videos and more.

Mistral AI Launches Text-to-Speech Model

info@uweb3.io
-
0
Mistral AI Launches Text-to-Speech Model

Mistral AI is expanding its Voxtral model family with its first text-to-speech model.

The launch comes amid intensifying competition in the fast-growing AI voice market, with Voxtral TTS pitched as an alternative to models from competitors including OpenAI and ElevenLabs.

The Paris-based startup unveiled its new system on Thursday. The 4 billion parameter model is designed for enterprise deployment across voice assistants, customer support and sales engagement tools. 

Unlike many rival offerings, Voxtral TTS has been released with open weights, allowing organizations to run the model on their own infrastructure rather than relying on third-party APIs.

The model supports nine languages: English, French, German, Spanish, Dutch, Portuguese, Italian, Hindi and Arabic.

Mistral said the model is lightweight enough to operate on consumer hardware, including laptops, smartphones and edge devices, while maintaining what it describes as “frontier-quality” performance. The company positions this as a key differentiator for enterprises seeking greater control over data, cost and customization.

Related:Cohere Unveils Open Source Speech Model for Edge Devices

 

Another key feature, Mistral said, is voice adaptability. The model can replicate a speaker’s voice using just a few seconds of reference audio, capturing not only tone but also accent, intonation and emotion.

“Our model excels at both contextual understanding and speaker modeling: capturing how a specific person naturally speaks,” Mistral wrote in a blog post. “With its compact size, low cost and latency and easy adaptability, Voxtral TTS gives full control and customization for enterprises looking to own their voice AI stack.”

Voxtral TTS can also perform cross-language voice control, such as generating English speech with a French accent, based on a short prompt.

In human evaluations of Voxtral, Mistral said its system matched or outperformed competing systems in terms of naturalness, exceeding lower-latency models from ElevenLabs while achieving parity with more advanced offerings in lifelike interaction.

The launch builds on Mistral’s earlier release of speech-to-text models and signals a broader push toward multimodal AI systems. 

1...652653654...1,326Page 653 of 1,326