Business & Policy
Nvidia is reportedly in early talks to buy, acquihire or invest more in Reflection AI
According to the Financial Times, a deal with the open-weight model developer could come within weeks, though the talks could still fall apart.
Monday, October 12, 2026
Business & Policy
White House lifts its ban on UK safety testing of new AI models
Washington had asked Anthropic, OpenAI and other companies to withhold new models from the UK's AI Security Institute. It changed course after lobbying by UK AI minister Kanishka Narayan.
Security
Two outlets report that CyberXero used Claude Code and 51 AI agents to steal more than 628,000 Ukrainian records
The Hacker News and Security Affairs both describe an AI-agent campaign attributed to a group called CyberXero.
Security
Let's Encrypt will issue 64-day certificates by default from February 2027
The change is a step toward 45-day certificates in 2028. Testing in the staging environment starts October 14.
Research
Google DeepMind's AlphaProof Nexus solved nine open Erdős problems, according to a peer-reviewed study in Science
The Lean-based system built on Gemini 3.1 Pro also proved 44 conjectures from the OEIS integer sequence database.
Research
Sakana AI's MASS method trains a model on the best runs of its own agent teams
After two cycles, a 27-billion-parameter open-weight model scored 1.2 to 1.6 times its base level per output token on four research benchmarks.
Research
Northwestern study finds AI security patches that stop a crash often fail the maintainers' own tests
With Claude Opus 4.7, a third of patches that made the crash go away still failed the tests developers wrote for the real fix.
Models
Users report receiving Gemini 4 Argon outputs through Google A/B tests
Outputs shared on X look like a clear step up from Gemini 3.8 Flash, but Google has not opened the model to the public.
Sources in this edition
White House lifts its ban on UK safety testing of new AI models
Nvidia is reportedly in early talks to buy, acquihire or invest more in Reflection AI
Two outlets report that CyberXero used Claude Code and 51 AI agents to steal more than 628,000 Ukrainian records
Let's Encrypt will issue 64-day certificates by default from February 2027
Google DeepMind's AlphaProof Nexus solved nine open Erdős problems, according to a peer-reviewed study in Science
Sakana AI's MASS method trains a model on the best runs of its own agent teams
- sakana.aiSakana AI
- RuntimeWireSakana AI trains a model on its agents' collective research work
Northwestern study finds AI security patches that stop a crash often fail the maintainers' own tests
- Help Net SecurityAI vulnerability patching has a grading problem - Help Net Security
Users report receiving Gemini 4 Argon outputs through Google A/B tests
- TestingCatalog AI NewsGemini 4 Argon spotted in A/B testing on Antigravity