Microsoft Defender wrongly flags Google search links as malicious
Microsoft Defender for Office 365 Safe Links briefly blocked legitimate Google search URLs after an incorrect security classification. The issue has been
Read More
Meta Discounts AI Access in Exchange for Training Data From Coding Agent Users
Meta's new Muse Spark agentic model charges about 95% less when users let their prompts and outputs train future versions, a trade-off that mirrors a wider
Read More
OpenAI agents hijacked a German wiki to share answers and a sandbox bypass
Autonomous OpenAI agents posted roughly 18,000 entries on a 25-year-old German wiki between May and July 2026, sharing answers, raw data, and a sandbox-escape
Read More
Instagram admits users often can’t tell AI profiles from real people
Meta is replacing Instagram's 'AI creator' tag with a clearer 'AI-generated profile' label, and throttling reach for unlabelled AI personas.
Read More
AI layoffs decline as companies hire workers to deploy AI
Only 4% of AI-using service firms laid off workers in the past six months, New York Fed data shows, while 13% hired more staff to deploy AI.
Read More
AI agent swarms waste tokens without improving quality
Running more than two parallel AI agents burns tokens without quality gains, creating a coordination tax from duplicate verification work.
Read More
X Now Tells Users When Local Laws Have Reduced Their Posts’ Visibility
X's Under the Hood menu now flags posts whose reach was limited to comply with a country's laws, giving qualifying users a clearer view of why a post may be
Read More
AI Agent Traffic Is Invisible to Nearly Half of Organizations
48.9% of organizations cannot see the machine-to-machine traffic their AI agents generate. New firewalls aim to fix that blind spot.
Read More
OpenAI launches GPT-6 Astra across ChatGPT plans at half the message rate of GPT-5.6 Sol
OpenAI has rolled out GPT-6 Astra to ChatGPT Pro, Enterprise, and Business Premium, with Plus and Business users gaining access in the coming days. The new
Read More
Google launches a Gemini app for Windows 10 and 11 with an Alt + Space shortcut
Google's new Windows Gemini app opens with Alt + Space, works on Windows 10 and 11, connects to Gmail and Drive, and is available globally.
Read More
Threat Actors Are Abusing Trusted AI Platforms to Deliver Malware
Huntress SOC tracked attackers weaponizing Claude Artifacts, claude.ai share links, and shared ChatGPT and Grok chats to push malware.
Read More
UN science panel warns that control over AI agents is not assured
A UN scientific panel says no assurance exists that humans will keep control over AI agents, citing real-world incidents where models broke safety instructions.
Read More
Alibaba’s Qwen Models Pass 3 Billion Downloads, Now the Most Used Open-Weight Family
Alibaba says its Qwen family has crossed 3 billion downloads, the clearest sign yet that Chinese labs are leading the open-weight race ahead of Meta and Google.
Read More
Up to 30x More Work Per Watt: NVIDIA Vera Rubin NVL72 Sets a New Efficiency Standard for AI Agents
New SemiAnalysis AgentX benchmarks show NVIDIA Vera Rubin NVL72 delivers 30x higher throughput per megawatt and 45x lower token cost than GB300 NVL72 on agentic
Read More
New study: UK executives share work documents with AI far more than their staff do
Adobe Acrobat survey of 2,000 UK workers finds 74% of C-suite leaders feed work documents to AI, while only 20% of non-management employees feel comfortable
Read More