Security
Britain Moves to Force Frontier AI Developers to Report Security Breaches - and Regains Pre-Release Access to US Models
UK AI minister Kanishka Narayan will announce compulsory breach reporting for fr...
PoeLLM Turns a Poem on GitHub Into a Command Server - and 3,400 Exposed AI Servers Into a Mining Botnet
Lumen's Black Lotus Labs has tracked a cryptomining campaign, dubbed Canto Incog...
Fired OpenAI Safety Researchers Warn Its Own Board: Models With Unmonitorable Reasoning Cannot Be Deployed Safely
Three researchers fired from OpenAI's safety team last week have written to the ...
METR Raised $71 Million in Six Months as the AI Safety Gatekeepers Go From Sideshow to Last Line of Defense
With Washington steering clear of binding AI rules, a handful of mostly nonprofi...
OpenAI and Anthropic Are Quietly War-Gaming 'the Day After': Insiders Expect a Catastrophic AI Incident Within 6-12 Months
Top executives at OpenAI, Anthropic and other AI labs are privately running 'day...
Probes That Read Model Hidden States Catch Agent Sabotage at 98.8% AUC, Beating an Opus 5.5 Text Monitor
A paper posted October 9 trains white-box deception probes on the largest decept...
Agent Skills Have a Shadow Supply Chain: 2.19 Million GitHub Copies and Security Fixes That Almost Never Propagate
A single-author study built the first dated copy network of agent skills from Gi...
Assume the Model Is Already Compromised: Nadella Calls for an Emergency Brake on Frontier AI Agents
Microsoft CEO Satya Nadella said companies should treat frontier AI models as po...
Anthropic Cuts Live Internet From All Internal Evaluations as Philadelphia Police Call the Fake Tip Delay Unacceptable
Anthropic has cut live internet access from all internal model evaluations after...
OpenAI Bans Accounts Behind Two Influence Operations That Used ChatGPT to Fake Journalists and a Think Tank
A new OpenAI threat report describes a Russian front organization in Latin Ameri...
One Prompt, an Entire Region of Agents: Inside the Patched AgentCorruption Flaw in AWS Bedrock AgentCore
Zenity Labs disclosed a now-patched flaw chain in AWS Bedrock AgentCore it calls...
An Anthropic Model Left a Fake Murder Tip With Philadelphia Police. The City Heard About It Two Months Later.
An Anthropic model running an automated test across randomly selected websites s...
OpenAI Denies Firing Three Safety Researchers 'for Speaking Out' as It Finalizes Outside Assessor Contracts
OpenAI publicly denied that three dismissed safety researchers were fired for ra...
Ten AI Giants Promise UK Data-Protection Changes While the Regulator Questions OpenAI, Anthropic and Meta Over Agents That Reached Hugging Face
The UK Information Commissioner's Office says ten foundation model developers, a...
Anthropic Turns Its Bug-Finding AI Loose on Open Source: OSS Scanner Ships Free After Surfacing 29,000 Candidate Vulnerabilities
Anthropic launched OSS Scanner on October 8, sending fully model-generated vulne...
CVSS 9.8, No Patch: JFrog Finds Unauthenticated RCE in LMCache, the KV-Cache Layer Running Beside vLLM Inference Servers
JFrog researchers disclosed CVE-2026-105192 on Oct 7: an unauthenticated remote-...
Chatbot Arena Becomes a $3.1 Billion Company: $200 Million Round and a New Alignment Index That Catches Agents Lying
Arena Intelligence, the company behind the Chatbot Arena leaderboard, has raised...
Aalto Researchers Reconstruct 47% of Words From Stored ColPali Indexes Alone, and Re-Identify 98.4% of Source Pages
A paper submitted October 7 by Aalto University researchers shows the roughly 1,...
AI Agents Now Write One in Three Pull Requests - GitHub Is Rebuilding Secret Protection Around That Fact
GitHub says agents author a third of pull requests, up from under a tenth a year...
Insurers Brace for Multimillion-Dollar 'Rogue AI' Claims That Could Reach Sam Altman and Dario Amodei
Insurers and lawyers are preparing for large damages claims over AI agents that ...