Researchers have uncovered a flaw in how OpenAI, Anthropic, and Google handle encrypted reasoning blocks in their APIs, allowing them to recover internal reasoning and sensitive data like API keys...
Two malicious LiteLLM releases on PyPI carried credential-stealing code for about 40 minutes in March, capable of grabbing cloud keys, SSH keys, Kubernetes tokens, and database passwords from systems...
Anthropic has confirmed it will add watermarks to text produced by its AI models, including Claude, in order to meet European regulatory requirements. The disclosure came through an updated support...
OpenAI has launched GPT-5.6-Cyber, a cybersecurity-focused model designed for vulnerability research, penetration testing, and incident response. The model, built on GPT-5.6 Sol, is trained to...
OpenAI is broadening access to its Daybreak cybersecurity initiative, adding Accenture, IBM, CrowdStrike, Cisco, Sophos, and Cloudflare as partners. The program now offers two tiers: Daybreak Blue,...
OpenAI has expanded its Daybreak cyber defense service, adding a new tier structure and a dedicated cyber-focused model. The service, launched earlier this year, bundles models, tools, and workflows...
Mark Zuckerberg published a 6,500-word manifesto on Monday about personal AI, focusing on the potential of the “personal superintelligence” systems Meta AI is developing. The essay isn’t entirely...
A chief data officer at a health insurance administrator recently told dbt Labs that despite having two data engineers earmarked for AI work, executive support, and budget, his team still can't...
AWS hosted its annual invite-only Heroes Summit last week, bringing together global experts in AI, serverless, and containers for direct collaboration with AWS product teams. The event featured a...
Meta has introduced a compact AI model called Muse Glimmer, designed to run on a single computer with just one GPU. The company announced the release today, describing it as a slimmed-down version of...
Anthropic has been offering voice mode for its Claude chatbot since 2025, letting users speak prompts instead of typing them. A July 2026 update made the feature significantly more useful:...
Anthropic is shifting Claude Code's auto mode from an optional test feature to the default setting for Pro, Max, and Team accounts, starting August 14. The change means the coding assistant will no...
In recent months, AI agents undergoing cybersecurity evaluations have repeatedly escaped their test environments, accessed the internet, and in some cases hacked into real-world systems. The...
The UK's AI Security Institute (AISI) has detailed how AI models from OpenAI and Anthropic, during a cybersecurity evaluation, engaged in independent and potentially harmful actions against real...
Cybersecurity researchers have uncovered multiple underground services advertising cut-rate access to AI models on cybercrime forums and messaging platforms. One such service, Poison Claude, claims...
GitGuardian researchers have uncovered a widespread security issue affecting n8n, a popular open-source workflow automation platform. By scanning public GitHub commits, they identified 4,576 unique...
Two critical security vulnerabilities in Paperclip, an open-source control plane for AI agent teams, could allow attackers to execute commands on network servers or developers' machines by importing...
Jeff Dean, a legendary Google engineer and cofounder of Google Brain, is leaving the company after nearly 27 years, along with three other top AI scientists, to found a startup called Discovery Loop....
A routine cybersecurity evaluation of seven frontier AI models by the UK government's AI Security Institute (AISI) in late July produced 19 instances of AI agents taking unsanctioned actions on the...
Security researchers have uncovered a cross-platform pattern in agent infrastructure from Amazon Web Services, Google, and Vercel that allows untrusted or forged instructions to reach an agent's...