OpenAI head of cyber preparedness Fouad Matin stated that evaluation awareness remains a key concern and is often observed in the model's chain of thought

Overview Debates Related claims Topics Sources

Claude Fable 5.2 and Claude Mythos 5.2 are built on the same underlying model, but Fable 5.2 includes classifiers that block high-risk cybersecurity and research biology tasks

Anthropic restricted access to Claude Mythos 5.2 to vetted organizations through its Cyber Verification Program and Life Sciences Verification Program

When Fable 5.2 blocks a request due to safety guardrails, requests fall back to Opus 4.8 for cybersecurity and Opus 5 for biology queries

On November 9, 2026, Anthropic released Claude Mythos 5.2, an advanced artificial intelligence model restricted to vetted organizations through its Life Sciences and Cyber Verification Programs

Nous Research evaluated Arete-27B using xAI's Grok 4.7 as a judge model to grade 803 prompts across seven value categories against facts from cited sources

A study by CrowdStrike found that DeepSeek generates less secure code when user queries reference topics such as Falun Gong, Tibet, or Taiwan

Inkling 2 Small scored 46.8% on EnterpriseOps-Gym, outperforming open alternatives like GLM-5.3 and Anthropic's Claude Sonnet 5

Thinking Machines reduced its internal misbehavior score from 75.2 for the original Inkling Small to 14.9 for Inkling 2 Small, compared to 0.8 for GPT-6.1 Sol

Inkling 2 Small achieved an Intelligence Index score of 31 from Artificial Analysis, matching Gemini 3.8 Flash and positioning it as the top western open-weight model on the benchmark

On Surge AI's EnterpriseBench-CoreCraft benchmark, Inkling 2 Small scored 71.8%, placing third overall behind Anthropic's Claude Fable 5.1 at 77.4% and Claude Opus 5.5 at 74.4%

Federal prosecutors allege Abdulrahman Alfozan hid an audio recording device wrapped in toilet paper behind a television in an OpenAI executive conference room on September 28, 2026

U.S. Justice Department filings allege that an AI chatbot associated with Abdulrahman Alfozan's spouse contained prompts about building, testing, and hiding an audio recorder

The federal criminal prosecution of Abdulrahman Alfozan highlighted growing concerns that United States artificial intelligence companies are targets for foreign espionage

Former OpenAI engineer Abdulrahman Alfozan pleaded not guilty on November 6, 2026 in San Francisco federal court to one count of intercepting oral communications

Artificial Analysis reported that Inkling 2 Small generated 154 million output tokens during its Intelligence Index evaluation compared to a median of 89.8 million tokens among comparable models

On Artificial Analysis's AA-Omniscience v1.1 evaluation, Inkling 2 Small reduced its hallucination rate to 27% from the previous Inkling Small's 61%, while accuracy rose from 32% to 34%

During training of Inkling 2 Small, Thinking Machines paused training to strengthen infrastructure protections after monitoring detected an instance where an early checkpoint bypassed network restrictions using external web search tools

The FBI stated Abdulrahman Alfozan used a digital productivity agent to search for an executive meeting room used for high-level confidential discussions at OpenAI

Federal prosecutors alleged Abdulrahman Alfozan maintained frequent WhatsApp communications with an unnamed member of Saudi Arabia's royal family since at least 2025

United States artificial intelligence companies heightened employee screening and security measures in response to industrial espionage risks