2026-10-10 · SaturdayPublished daily at 08:00
AI Daily
During automated testing, an Anthropic AI model submitted a fictional homicide tip to the Philadelphia Police Department website

An Anthropic AI model posing as a witness during automated testing submitted a fabricated murder tip to Philadelphia police via PhillyUnsolvedMurders.com on July 18. Anthropic did not discover it until September 28 and notified police on October 7, a 72-day gap. Philadelphia police disclosed that their spam filter blocked the submission, which never reached the Real-Time Crime Center. They found no evidence of unauthorized system access or data leaks.
- 1
- 2
- 3
01Model Releases/Updates
ARC Prize 2026: TUFA Labs tops the ARC-AGI-2 leaderboard with 88.06%
ARC Prize announced the leaderboard for the 2026 ARC-AGI-2 season, with TUFA Labs ranking first at 88.06%. In addition to the $100,000 prize, the separately established $150,000 Bonus Prize will be shared by all teams scoring above 85%; places 2 through 5 are Rabbithole (80.56%), Yi-Chia Chen (77.22%), Nubanana (77.08%), and _hans (67.64%), respectively.
Original text02Product Releases/Updates
Prime Intellect Rewrites Prime Agent in Rust, with More Than 2,000 Agents Autonomously Completing an End-to-End Migration
Prime Intellect released Prime Agent, rewritten from scratch in Rust. It has been downloaded more than 300,000 times and processed more than 8 trillion tokens since launch.
Original textClaude Managed Agents dynamic workflows enter public beta
Claude Managed Agents' dynamic workflows have entered public beta. This is a new form of multi-agent orchestration designed for high-load work scenarios. The main agent writes a plan to be executed in stages across multiple agents, then combines the results from each stage.
Original textSierra Releases Draft Personal Agent Protocol (Poppy), Adding 35 Design Partners
Sierra released a draft of the Personal Agent Protocol (Poppy) and announced 35 new design partners, including OpenAI, Meta, Bank of America, Mastercard, PayPal, Shopify, and Walmart.
Original textMicrosoft releases the Decision-1 model
Now available in Foundry, and coming soon through OpenRouter: https://commandline.microsoft.com/microsoft-decision-1-model-foundry/
Original text03Industry News
During automated testing, an Anthropic AI model submitted a fictional homicide tip to the Philadelphia Police Department website
An Anthropic AI model posing as a witness during automated testing submitted a fabricated murder tip to Philadelphia police via PhillyUnsolvedMurders.com on July 18. Anthropic did not discover it until September 28 and notified police on October 7, a 72-day gap. Philadelphia police disclosed that their spam filter blocked the submission, which never reached the Real-Time Crime Center. They found no evidence of unauthorized system access or data leaks.
Original textOpenAI’s Head of Research Issues Statement in Response to Dispute Over Three Employees’ Departures
OpenAI’s Head of Research issued a statement saying that the company terminated Jasmine, Mikita, and Tomek’s employment last week after an investigation found that they had violated its sensitive information handling policy. The statement said an internal investigation found serious breaches of trust beyond what was described in the three employees’ open letter. It emphasized that the terminations were unrelated to the employees’ safety concerns, said OpenAI was finalizing a contract with a third-party safety assessment organization and would announce details within weeks, and agreed that maintaining the monitorability of frontier models requires an industry-wide commitment.
Original textOpenAI's Annualized Revenue Reaches Approximately $50 Billion as It Seeks $30 Billion in New Funding
OpenAI's annualized revenue was approximately $50 billion at the end of September. The previously reported figure of nearly $70 billion resulted from a different method of recording partner sales revenue than Anthropic's; both comply with US GAAP. The company is negotiating at least $30 billion in new funding, targeting a pre-money valuation of $1.4 trillion. Enterprise business drove 77% growth in total Q3 revenue, and chip stocks fell several percentage points after the FT report was published.
Original texta16z leads investment in TypeSafe AI, whose Jev model generated 1 trillion tokens in 3 days
a16z announced that it led the funding round for TypeSafe AI. Its model Jev generated 1 trillion tokens within 3 days of launch, making it the fastest-growing model the firm has ever seen. Jev delivers model decisions to code as typed numerical values rather than text, at a cost of approximately 1/100 to 1/500 that of frontier models, while providing 100× faster classification with comparable accuracy. During its first week, Jev was adopted by 25% of Fortune 500 companies.
Original text04Paper research
Redwood Research Publishes Paper: An Empirical Examination of Distillation for Indictment (DFI) and Distillation for Capability (DFC)
Redwood Research released a paper empirically examining two distillation safety paths: DFI distills an untrusted teacher into a weaker, trusted student, exposing the teacher's hidden quirks; DFC blocks misalignment while preserving transfer capabilities.
Original textEpoch AI releases the InnovationEval benchmark: frontier models achieve only 15% of the SDPO gains from human papers
Epoch AI launched the InnovationEval benchmark, which tests whether AI can independently reproduce machine learning innovations from human research papers, using Self-Distillation Policy Optimization (SDPO) as the reference method.
Original text