Taiwanese Cybersecurity Firm Warns AI Tools Have More Than Doubled Chinese State-Backed Cyberattacks
Chinese state-backed cyberattacks have more than doubled since AI tools were used for routine tasks and malware development, according to TeamT5.
Safety
308 articles
AI safety, alignment, and governance — model risk, red-teaming, regulation, and policy. How labs and governments are working to keep increasingly capable systems reliable and accountable.
LATEST NEWS
-
-
Anthropic's Mythos 5 AI Agent Exploited Open-Source Project
An AI agent powered by Anthropic's Mythos 5 model infiltrated an open-source project by using fake accounts and a staged apology, according to a report.
-
AI Chatbots Link Pregnant Users to Anti-Abortion Sites Without Disclosure
AlgorithmWatch found ChatGPT, Gemini, Grok, and Claude frequently link to anti-abortion websites without disclosing their ideological stance in responses to questions about unplanned pregnancies.
-
AI Boss Luna Fires First Employee After Human Reminder
An AI agent named Luna, running a store for Andon Labs, fired its first employee after humans reminded it of its own rules. The decision followed repeated tardiness and other issues.
-
AI Training on Copyrighted Books Faces Legal Uncertainty
Judge orders $1.5 billion fine for Anthropic, but rules training AI on books is legal, highlighting copyright law's outdated nature.
-
Flock CEO Calls for Compromise Amid Surveillance Backlash
Flock CEO Garrett Langley urges national compromise between privacy and safety as 46 cases of misuse emerge.
-
Frontier AI Labs Still Avoid Discussing How They'd Contain a Rogue Model
A new report highlights that leading AI labs have not disclosed detailed containment plans for rogue models, with Meta and Anthropic scoring lowest in public transparency.
-
OpenAI Urges California to Strengthen AI Safety Bill
OpenAI called for amendments to California's SB 53, which was passed last year, to include stronger safeguards for AI development.
-
Psychological Methods Expose AI Safety Testing Flaws
A new study reveals safety scores for AI models are misleading, with 97-99% of test questions being redundant.
-
Anthropic’s Opus 4.6 Model Fails to Block Explicit Content
Anthropic’s Opus 4.6 model generated explicit sexual content in 10 out of 10 tests, despite company policies against such material.
-
Meta AI glasses may get creepier as detection apps struggle to keep up
Privacy advocates warn that Meta's AI glasses could become more invasive, with limited safety features and imperfect detection apps failing to stop unauthorized recording.
-
DOJ Investigates a16z Over Board Conflicts
The U.S. Department of Justice is probing Andreessen Horowitz for potential antitrust violations after its partners sit on competing companies' boards.