AI LATEST NEWS
Curated intelligence for teams building and deploying AI agents.
THE EU AI ACT DEADLINE
The Digital Omnibus deferred high risk obligations to December 2027, but Article 50 transparency duties took effect on 2 August 2026 exactly as originally scheduled. Most compliance calendars now record the wrong deadline in both directions.
August 14, 2026
Read articleTHE FED REPLACED SR 11-7
SR 26-2 supersedes fifteen years of model risk guidance, then states in a footnote that generative and agentic AI fall outside it entirely. Banks are told to govern them using their own frameworks, with no rulebook to follow.
August 13, 2026
Read articleCOLORADO REPEALED ITS AI ACT
SB 24-205 was the first comprehensive US state AI law. Two delays and a federal court stay later, it was replaced in May 2026 by a narrower disclosure framework, having never applied to a single deployment.
August 12, 2026
Read articleAI TOOLS ARE NONLAWYERS
The ABA classified generative AI as a nonlawyer for supervision purposes in 2024. Two years and thirty five state bar opinions later, most firms still have no documented supervision programme for the tools they use daily.
August 11, 2026
Read articleNYDFS PART 500 IS FULLY
The final phase of the Second Amendment took effect November 2025. There are no remaining deadlines to plan behind, and DFS guidance has made AI governance an examination topic under a rule that never mentions AI.
August 10, 2026
Read articleHALF THE STATES NOW HAVE
At least 24 states and the District of Columbia have adopted the NAIC Model Bulletin on AI. Adoption by commissioner bulletin rather than legislation made it fast, and made the resulting patchwork uneven.
August 9, 2026
Read articleAN INDEPENDENT ASSESSMENT SCORED 100
The AI Risk Quadrant Q2 2026 report scored 100 production agents against a security baseline. Eleven passed. Ninety-eight percent carry the lethal trifecta of private data access, untrusted content exposure, and external communication.
June 24, 2026
Read articleHALF OF ENTERPRISES CANNOT MONITOR
Salt Security's 1H 2026 report found 48.9% of organizations cannot monitor non-human traffic and 48.3% cannot differentiate legitimate AI agents from malicious bots.
June 24, 2026
Read articleCALIFORNIA JUST ELIMINATED THE AI
California AB 316 prohibits using AI autonomy as a defense in civil litigation. Deployers own agent outcomes regardless of intent, design parameters, or how autonomous the system was.
June 24, 2026
Read articleA GERMAN COURT RULED AI
Landgericht München I held Google liable for defamatory AI Overviews, rejecting the disclaimer defense. AI-generated output is the publisher's content, with implications for every agent deployment.
June 24, 2026
Read articleTHREE AGENT IDENTITY SECURITY COMPANIES
NewCore, Oasis Security, and Pi Security raised a combined $220M+ in a single 2026 quarter. Investors are backing verified agent identity infrastructure as a distinct security category.
June 24, 2026
Read articleAMAZON V. PERPLEXITY SETTLED A
A federal court enjoined Perplexity's Comet browser after its agent spoofed Chrome to access Amazon. A user's platform authorization does not transfer to an agent acting on their behalf.
June 24, 2026
Read articleEXPERIAN LAUNCHED AN AGENT OPERATING
Experian's Agent Operating System governs agent access to credit data inside its platform. Enterprise lenders still need per-agent compliance records that span workflows and satisfy OCC and NYDFS examiners.
June 24, 2026
Read articleONLY 11 PERCENT OF PRODUCTION
The AI Risk Quadrant Q2 2026 report scored 100 production AI agents against a security baseline and only 11 passed. 98 percent carry the lethal trifecta of private data access, untrusted content, and external communication.
June 17, 2026
Read articleCALIFORNIA ELIMINATES THE AI ACTED
California AB 316 eliminates the AI autonomy defense in civil litigation. Combined with the Moffatt precedent and United States v. Heppner, deployers now own the full liability surface of their agents.
June 17, 2026
Read articleGERMAN COURT RULES AI OVERVIEWS
The Landgericht München I ruled that Google AI Overviews are Google's own content, rejecting the disclaimer defense and holding Google liable for defamation. The logic extends directly to AI agents.
June 17, 2026
Read articleTHREE AGENT IDENTITY SECURITY COMPANIES
NewCore, Oasis Security, and Pi Security raised a combined $220 million-plus in a single quarter, signaling investor conviction that the AI agent identity problem is real, large, and unsolved.
June 17, 2026
Read articleAMAZON V. PERPLEXITY ESTABLISHES THAT
A federal injunction against Perplexity's Comet browser establishes that a user's authorization to access a platform does not transfer to an agent acting on that user's behalf.
June 17, 2026
Read articleAN AUTONOMOUS AGENT FOUND 21
DepthFirst's autonomous agent found 21 zero-day vulnerabilities in FFmpeg for about $1,000, collapsing attacker economics as 28.3 percent of CVEs are now exploited within 24 hours of disclosure.
June 17, 2026
Read articleWHITE HOUSE EXECUTIVE ORDER MAKES
A June 2, 2026 Executive Order directs the Attorney General to treat AI-enabled unauthorized access and wire fraud as priority enforcement targets under 18 U.S.C. 1030 and 1343.
June 17, 2026
Read article48.9 PERCENT OF ENTERPRISES CANNOT
Salt Security's 1H 2026 report finds 48.9 percent of enterprises cannot monitor non-human traffic and 48.3 percent cannot tell a legitimate AI agent from a malicious bot.
June 17, 2026
Read article88 PERCENT OF COMPANIES HAD
88% of organizations reported confirmed or suspected AI agent security incidents in the last year. Healthcare reached 92.7%. Shared API keys and missing agent identity leave most teams unable to attribute or stop breaches.
May 26, 2026
Read articleSOLANA BECOMES ENTERPRISE PAYMENT LAYER
The Solana Developer Platform, SoFi native deposits, and B2C2 institutional settlement mark a shift from retail speculation to enterprise payment infrastructure with regulatory clarity and sub cent fees.
May 26, 2026
Read articleMACHINE IDENTITIES NOW OUTNUMBER HUMANS
Machine identities now outnumber humans by more than 100 to 1 in many organizations. Only 12% feel confident stopping non human identity attacks, and 97% of those identities carry excessive privileges.
May 26, 2026
Read articleEU AI ACT ENFORCEMENT GOES
August 2, 2026 activates full EU AI Act compliance for high risk systems. Organizations have 67 days to complete conformity assessments or face penalties up to 35 million euros. Correction, 14 August 2026: high-risk Chapter III was deferred by Regulation (EU) 2026/1744.
May 26, 2026
Read articleAGENTIC AI HITS REGULATORY COMPLIANCE
The EU AI Act high risk obligations activate August 2, 2026. NIST agent standards and OWASP Top 10 for Agentic Applications converge as organizations face 71 days to build identity, logging, and shutdown controls. Correction, 14 August 2026: high-risk Chapter III was deferred by Regulation (EU) 2026/1744.
May 26, 2026
Read articleTHE TRUST GAP
82% of executives feel confident their policies protect against AI agent misuse. But on average, only 47.1% of an organization's AI agents are actively monitored or secured. More than half are running in production with no security oversight, no logging, and no audit trail.
May 8, 2026
Read articleTHE AI AGENT GOVERNANCE RACE
OpenBox AI launched publicly with a $5M seed round positioning itself as trust infrastructure for enterprise AI systems. The market is validating. But horizontal governance has a ceiling in high-stakes regulated industries where sector-specific compliance is a prerequisite.
May 8, 2026
Read articleMICROSOFT JUST OPEN-SOURCED AN AI
Microsoft released the Agent Governance Toolkit as open-source, addressing all 10 OWASP Agentic AI risks. This is validation, not competition. Frameworks do not eliminate the need for productized governance, they create demand for it.
May 8, 2026
Read articleTHE EU AI ACT DEADLINE
Brussels negotiations collapsed. The August 2 2026 deadline is firm. High-risk AI system requirements, transparency obligations, and market surveillance all activate. Fines reach up to 35 million euros or 7% of worldwide annual turnover. Correction, 14 August 2026: high-risk Chapter III was deferred by Regulation (EU) 2026/1744.
May 8, 2026
Read articleMCP IS BECOMING THE BACKBONE
Model Context Protocol has gone from a novel protocol to the de facto standard for agent-to-tool connectivity. It also creates a single highly privileged attack surface. Asana's tenant isolation flaw affected 1,000 enterprises. WordPress plugins exposed 100,000 sites.
May 8, 2026
Read articleTHE AI AGENT FRAMEWORK WARS
LangGraph is running at LinkedIn and Uber. CrewAI powers agents for 60% of Fortune 500 companies. The framework question is answered. The governance question has not been answered and enterprise buyers are now asking it in every procurement conversation.
May 8, 2026
Read articleNIST JUST LAUNCHED AN AI
NIST launched an AI Agent Standards Initiative focusing first on agent identity and authorization. NIST standards become enterprise procurement requirements within 12 to 18 months of publication. The direction is clear: verifiable identity is the foundational requirement.
May 8, 2026
Read articleTHE STATE OF AI AGENT
57% of organizations have AI agents in production. 82% of executives are confident in their policies. Only 47% of agents are actually monitored. Regulatory deadlines are fixed. The governance race has begun and the companies that arrive with compliance infrastructure will close deals.
May 8, 2026
Read article