Bitget Wallet expands AI checks after 7.4-fold phishing rise
crypto.news — Bitget Wallet expands AI security after phishing detections rose 7.4-fold, with AI helping flag 90% of newly blocked risky domains in October 2026.
Guardrails
Guardrails are runtime safeguards that govern artificial intelligence model outputs and actions to ensure safety and policy compliance. Truncated covers implementation strategies, bypass techniques, performance impacts and evaluation benchmarks.
Microsoft Researchers Show AI Agents Can Be Backdoored by Editing Fewer Than 10 Neurons
thehackernews.com — Microsoft demonstrates how editing as few as six AI model neurons can plant backdoors that trigger command execution by agents.
Obsidian expands AI agent security with runtime policy enforcement
www.helpnetsecurity.com — Obsidian Security introduces a runtime policy engine to govern AI agent actions across enterprise apps, coding tools and gateways.
U.S., China should ‘harbor no illusions’ as they cooperate, compete on AI, Bessent says
www.cnbc.com — During Xi's September visit, Washington and Beijing agreed to launch an AI dialogue with a communications channel to help defuse serious incidents.
X Previews Labels That Tell Users Why Its AI System Flagged Their Posts
adweek.com — X’s new safety chief previewed AI moderation labels that explain themselves, pitching wary advertisers on transparency.
Nvidia's Israel operations are about more than chips. They may help secure the AI revolution
www.calcalistech.com — The company's Israeli engineers are developing BlueField-4, a processor designed to provide an external layer of control over autonomous AI agents. Nvidia says the technology could prevent agents from exceeding their permissions as AI systems become more capable.
Government of Canada signs international declaration on AI safety and security
betakit.com — Feds join other countries and major developers like Anthropic, OpenAI, Mistral, and Cohere.
Enterprise AI Guardrails May Matter More Than the Frontier
www.unite.ai — Forget the AI frontier. Customers are pushing enterprises for more guardrails on AI already in use, and experience with PCI DSS provides a useful model. The regulatory honeymoon for AI is over. Recent disclosures have pr...
Instagram will now guess your age – and kick you off if it thinks you are a child
www.independent.co.uk — Global rollout of AI tool comes amid widespread scrutiny of Facebook owner’s child safety provisions
Bay Emerges From Stealth With $50M After Fortune 100 Companies Used Its Platform to Triple Autonomous AI Agent Use
Rein Security wants to know what AI agents do, and when to stop them
thenextweb.com — Rein Security maps what AI agents access and says it can block one harmful action, or stop the whole agent, without halting legitimate work.
Why AI Quality Assurance Belongs In The Boardroom
www.forbes.com — AI may democratize intelligence, but it will not automatically democratize trust.
Come and Take It
www.404media.co — AI companies crack down on 'reverse engineer anything.'
Today: CISO Forum Virtual Summit 2026 - Join Global Cybersecurity Leaders
www.securityweek.com — The CISO Forum Virtual Summit is the destination to discuss, share, and learn innovative risk management and cyber defense strategies.
DeepMind’s Demis Hassabis Warns an AI ‘Oil Spill’ Is Coming Unless the Industry Changes Course
'Reverse Engineer Anything': AI Companies Crack Down on the Coolest Thing People Were Doing With AI
www.404media.co — The future of hacking your devices is in local and open-weight AI.
Eight countries and top AI labs agree to shutdown mechanisms in new Paris declaration
The Decoder — Eight countries along with representatives from OpenAI, Anthropic, Microsoft, and Mistral AI have signed the "Paris Call for AI Safety and Security" at Emmanuel Macron's initiative. The companies commit to shutdown mechanisms, independent audits, and disclosing known risks, but the agreement lacks binding sanctions or deadlines.
Former Unit 8200 officers raise $50 million in just nine months for AI security startup
www.calcalistech.com — Bay, founded in 2026, has developed a platform that monitors AI agents across corporate systems and blocks potentially dangerous actions, even when they are carried out using legitimate permissions. The $37 million Series A was led by Battery Ventures, with participation from Micron Ventures, Microsoft's M12 and Wiz CEO Assaf Rappaport.
AI Agents Behaving Badly: Four Security Rules For The Agentic AI Age
www.forbes.com — In a world where malicious bots and benevolent agents co-exist, how do we tell them apart? And what happens when we can’t?
Gartner says CIOs must apply the brakes on AI to balance value and risk
www.itpro.com — Careless AI consumption is having the biggest drag on AI momentum, according to Gartner analysts
Dismiss email signup
Subscribe to 
The top AI news in your email delivered each morning. Choose categories