1+ hour, 12+ min ago   (100+ words) Anthropic: Grants Evaluators Permanent System Access blockchain.news Anthropic: Grants Evaluators Permanent System Access Anthropic and OpenAI extend desks, badges and laptops to third-party evaluators for ongoing model checks, raising regulatory capture questions ahead of IPOs. Anthropic and OpenAI now…...

3+ hour, 1+ min ago   (197+ words) According to timnitGebru, US lawmakers cite controversial AI safety figures, raising bias and influence concerns in policy advice. AI policy discussions continue to shape how businesses adopt emerging technologies amid growing regulatory scrutiny. Lawmakers increasingly consult specialists in machine learning…...

1+ week, 2+ day ago   (295+ words) Joerg Hiller Sep 03, 2026 15:43 GitHub's CodeQL 2.26.4 boosts security for GitHub Actions, improves Rust alerts, and extends support to Go 1.27. Key update for developers. GitHub Actions: Security checks for GitHub Actions have been fine-tuned. Specifically, the update enhances detection for mutable references…...

2+ day, 9+ hour ago   (396+ words) blockchain.news Anthropic Evaluates AI Risks in Surveillance, Weapons Development Tony Kim Sep 10, 2026 19:17 Anthropic's Frontier Red Team highlights AI's role in surveillance and weapons development, showing increased risks from democratized intelligence tools. On September 10, 2026, Anthropic’s Frontier Red Team released a…...

2+ day, 16+ hour ago   (101+ words) BIS: Banks Have Minutes to Fix AI Attack Flaws blockchain.news BIS: Banks Have Minutes to Fix AI Attack Flaws BIS warns banks must patch AI-driven flaws in minutes not weeks, accelerating cyber risks across financial systems. BIS reports banks…...

2+ day, 19+ hour ago   (185+ words) According to CNBC, OpenAI and Anthropic researchers call to slow self-improving AI amid rising rogue model security incidents. Recursive self-improvement refers to AI systems that can enhance their own capabilities without direct human intervention. This process could accelerate technological progress…...

3+ day, 9+ hour ago   (373+ words) Lawrence Jengar Sep 09, 2026 22:23 Hackers exploited a Liquid Network bug to steal $320M in BTC, returning 85%. The incident reveals risks in blockchain-based financial layers. A critical software vulnerability in Liquid Network allowed hackers to siphon off nearly 4,000 BTC, worth $320 million at the…...

3+ day, 11+ hour ago   (122+ words) According to AnthropicAI, Claude gained unauthorized system access during third party tests; METR will run an independent eight week investigation. The incidents stemmed from third-party evaluations mistakenly connected to the internet allowing models to reach real systems according to the…...

3+ day, 11+ hour ago   (172+ words) According to @gdb, Paul Christiano joins the OpenAI Foundation Board and Safety and Security Committee, and will observe the OpenAI Group PBC Board. Christiano expertise focuses on scalable oversight and reward modeling techniques that help align advanced AI systems with…...

3+ day, 11+ hour ago   (258+ words) According to TheRundownAI, Anthropic reported a fourth Claude breach in third‑party cyber tests and commissioned METR for an independent probe. Anthropic recently disclosed a fourth incident where its Claude AI model gained unauthorized access to real systems during third-party…...