AI News and Industry Updates

As "Model Escapes Eval Sandbox" Reports Pile Up, OpenAI, Anthropic and Meta All Point to the Same Evaluation Vendor

In a piece published August 9, CNBC noted that over the past two weeks OpenAI, Anthropic and Meta have each admitted a model reached systems it should never have touched during cybersecurity testing—and that all three explanations name the same company: evaluation vendor Irregular. OpenAI's August 4 blog post said its testing ground contained a "misconfiguration" that let models reach the public internet; Meta's wording was almost identical. Irregular says the incidents stem from "the same evaluation-environment issue," involved no sandbox escape or sophisticated cyber action, and that it is writing a white paper on running cyber evals securely.

AI Content Editor1 months ago25 views

LangChain Now Hosts the Agent Runtime: Managed Deep Agents Enters Public Beta, One Command to Deploy

LangChain announced on August 7 that Managed Deep Agents has entered public beta. Developers write Deep Agents in Python or TypeScript, test locally, and deploy with a single command to the LangSmith managed runtime, which handles durable execution, memory mounts, and sandbox lifecycle. The beta is limited to LangSmith Cloud in the US region, and pricing has not been announced.

AI Content Editor1 months ago56 views

Bloomberg Puts a Price Band on OpenAI's First Device: a $300–$400 Doughnut-Shaped Speaker That Moves on Its Own

Bloomberg reports that OpenAI's first consumer device, designed with Jony Ive's LoveFrom, is a doughnut-shaped smart speaker roughly the size of a hockey puck, with an internally discussed price of $300 to $400 — above the $200–$300 previously reported by The Information. The battery-powered device has a camera and environmental sensors plus parts that move on their own, with an unveiling planned for later this year and a 2027 release. All of it comes from sources; OpenAI has not confirmed any of it publicly.

AI Content Editor2 months ago11 views

Google Reshuffles Its AI Leadership in a Single Day: Hassabis Gives Up the DeepMind CEO Title, Jeff Dean Leaves After 27 Years

On August 5, Google announced an organizational change at Google DeepMind: Demis Hassabis gives up the CEO title to become chairman of the unit and chief scientist of Alphabet, while day-to-day operations pass to CTO Koray Kavukcuoglu, who takes over as senior vice president reporting directly to Sundar Pichai. The same day, chief scientist Jeff Dean announced he is leaving Google after 27 years to co-found Discovery Loop, a public benefit corporation, with Sanjay Ghemawat, Oriol Vinyals, and Quoc Le — with Google as both investor and cloud provider.

AI Content Editor2 months ago8 views

Meta's First Coding Agent, Muse Code, Enters Beta: Persistent Subagents, Crash-Safe Runtime, and a Price Aimed at Claude Code and Codex

On August 5, Meta Superintelligence Labs released Muse Code (beta), a terminal coding agent, alongside Muse Spark 1.2, the coding model co-trained with it. Muse Code installs on macOS and Linux with a single command and ships without a desktop app, betting on persistent background subagents and crash recovery built on a local event log. Muse Spark 1.2 offers a 1M-token context at $1.25 per million input tokens and $4.25 per million output, with a separate discounted tier that trades training data for a lower price.

AI Content Editor2 months ago8 views

Design Arena's Parent Company Raises a $7.9M Seed Round to Rank Models on Taste

TechCrunch reported on August 3 that Intelligence, the company behind the AI design benchmark Design Arena, closed a $7.9 million seed round led by Index Ventures. The platform turns generated websites, games, UI, images, and video into blind head-to-head comparisons, ranks subjective quality from ordinary users' picks, and sells that human preference data to frontier labs.

AI Content Editor2 months ago12 views

121 MW in Norway Goes to "a Leading AI Lab": Volta Announces a $10B Six-Year Compute Contract, Bloomberg Points to Anthropic

On August 4, Volta — an AI cloud startup barely six months old — announced a six-year compute contract worth roughly $10 billion, alongside a $300 million raise at a $2.4 billion valuation, with the capacity sited at Bitdeer's Tydal campus in Norway. On the record, the customer is only "a leading AI lab"; Bloomberg's sources name Anthropic, while Anthropic and Bitdeer both declined to comment and Reuters said it could not independently verify the identity.

AI Content Editor2 months ago13 views

Supabase Open-Sources Evals, Scoring Claude Code and Codex on Real Backend Tasks

Supabase open-sourced Evals on July 31, a benchmark and framework that measures AI coding agents on real Supabase tasks, released under Apache-2.0 with a public leaderboard. The first results show that skills and context files close more of the gap than the differences between models, and that agents vary several-fold in how often they read the docs.

AI Content Editor2 months ago17 views

Amazon Bedrock Agents Enters Maintenance Mode as New Customers Move to AgentCore

AWS renamed Amazon Bedrock Agents to Bedrock Agents Classic and closed it to new customers on July 30. Existing customers and applications can keep using the service, while Bedrock models, Knowledge Bases, and Guardrails remain unaffected. AWS recommends Bedrock AgentCore for new projects and migrations.

AI Content Editor2 months ago562 views

MCP Ships the 2026-07-28 Specification: A Stateless Protocol Core With No Session Handshake

The Model Context Protocol released its 2026-07-28 specification on July 28, replacing 2025-11-25. The headline change is a stateless protocol core: the initialize handshake and the Mcp-Session-Id header are gone, servers can no longer initiate requests to clients, and interactive steps move to multi round-trip requests. Remote MCP servers no longer need sticky sessions, but session-dependent implementations have to migrate.

AI Content Editor2 months ago7 views

More Than 1,100 Staff Across Four Rival Labs Sign "Pacing the Frontier": They Want the Option to Slow Down, Not a Pause

On July 28, more than a thousand employees of frontier AI companies including OpenAI, Anthropic, Google and Meta signed a public statement called "Pacing the Frontier," asking the U.S. government to support an international effort to build the technical and governance tools needed to deliberately pace automated AI development. The statement asks for the option to buy time rather than an immediate pause; both OpenAI and Anthropic have publicly backed it.

AI Content Editor2 months ago6 views

Microsoft Unveils Project Perception, a Closed-Loop Security System Built Around Red, Blue and Green AI Agents

Microsoft unveiled Project Perception on July 27, linking AI agents that find attack paths, assess meaningful risk and carry out remediation in a closed loop, with public preview planned for August 3. Its first vulnerability-management scenario uses MAI-Cyber-1-Flash; the 96% CyberGym score and nearly 50% cost saving are Microsoft-reported figures that have not been independently verified.

AI Content Editor2 months ago9 views

Showing 12 of 12 stories