A Developer Tool Compromise Reached OpenAI's App-Signing Workflow
OpenAI disclosed on 31 March 2026 that a compromised Axios package had affected part of its developer workflow, prompting certificate rotation.
How to Organise Your ChatGPT File Library
Clear names, approved data, and a regular cleanup turn ChatGPT's reusable files into a working space rather than a forgotten archive.
What 81,000 People Said They Wanted from AI
Anthropic's 18 March 2026 study gathered hopes and concerns from Claude users across languages and countries.
March Patch Tuesday Fixed Two Public Zero-Days
Microsoft's 10 March 2026 release addressed two publicly disclosed zero-days and serious Office issues.
GPT-5.3 Instant Focused on Clearer Answers and Fewer Dead Ends
OpenAI released GPT-5.3 Instant on 3 March 2026, updating ChatGPT's everyday model with changes to accuracy, tone, and refusal behaviour.
OpenAI's Threat Report Showed AI Inside Familiar Criminal Workflows
OpenAI's 25 February 2026 report described disrupted accounts linked to scams, cyber activity, influence work, and other abuse.
Claude Code Security Put AI to Work Finding Vulnerabilities
Anthropic announced a limited research preview on 20 February 2026 that reviewed code for vulnerabilities and suggested patches.
February's Windows Update Patched Six Exploited Zero-Days
Microsoft's 10 February 2026 security release addressed six vulnerabilities that were already being exploited.
Malicious OpenClaw Skills Turned a Viral Agent into a Security Warning
Koi Security reported 341 malicious skills posing as useful OpenClaw add-ons, showing why agent extensions need software-level scrutiny.