Artificial intelligence news.
Straight from the original announcements.

Claude Mythos

CSET explains how black-box distillation works and how Chinese labs may use it
ResearchOctober 26, 2026, 14:41

CSET explains how black-box distillation works and how Chinese labs may use it

Georgetown's Center for Security and Emerging Technology (CSET) has published an explainer on black-box distillation, the technique of training a weaker "student" AI…

Cognition upgrades Devin Security with Claude Mythos deep scans and big credit pledge
Safety and ethicsOctober 26, 2026, 17:00

Cognition upgrades Devin Security with Claude Mythos deep scans and big credit pledge

Cognition has launched a major upgrade to Devin Security, its product for finding, prioritizing and fixing vulnerabilities, adding deep scans powered by Claude Mythos…

Safety and ethics
Safety and ethicsOctober 19, 2026, 07:00

HPE warns AI finds software flaws faster than firms can patch them

AI is closing the gap between finding a software flaw and exploiting it, and enterprises need to move their security processes to machine speed, according to a blog post…

Anthropic's Free OSS Scanner Shifts the Security Bottleneck to Remediation
Safety and ethicsOctober 16, 2026, 21:49

Anthropic's Free OSS Scanner Shifts the Security Bottleneck to Remediation

Anthropic has launched OSS Scanner, a free, opt-in vulnerability scanning service for eligible open-source projects. According to JFrog's analysis, the tool makes…

Lean SuperIntelligence Debuts Self-Improving On-Prem Cybersecurity LLMs
Safety and ethicsOctober 9, 2026, 19:39

Lean SuperIntelligence Debuts Self-Improving On-Prem Cybersecurity LLMs

Lean SuperIntelligence (LSI), a Palo Alto startup, has closed a pre-seed financing round and introduced two cybersecurity language models, LSI-Offense-1 and…