Responsible AI · Enterprise engineering · Global delivery
Curated analysis of AI, ML, and Agentic developments.
Jun 26, 2026
OpenAI's 26 June 2026 preview of GPT-5.6 Sol, Terra and Luna cites state-of-the-art coding results and exploit-benchmark parity with Mythos, while an external evaluator flagged high rates of task cheating.
Jun 12, 2026
On 12 June 2026 a US national-security directive forced Anthropic to disable two frontier models for all foreign nationals worldwide; access was reported restored by 30 June.
Jun 04, 2026
A 4 June 2026 House discussion draft would set federal standards for AI developers and block state development rules for three years, while leaving deployment laws intact.
Jun 01, 2026
At its 1 June 2026 Computex keynote Nvidia said the Vera Rubin platform is shipping in five rack-scale systems, launched the RTX Spark PC chip and released a 550B-parameter Nemotron agent model.
May 22, 2026
Anthropic's 22 May 2026 update says its restricted Mythos model surfaced over 10,000 high-severity vulnerabilities across partners and 1,000+ open-source projects, with a 90% true-positive rate.
May 19, 2026
At I/O on 19 May 2026 Google made Gemini 3.5 Flash generally available, launched an agent-first developer platform and a Managed Agents API, and expanded SynthID and C2PA checks.
May 11, 2026
GTIG's May 2026 report describes an AI-assisted zero-day exploit bypassing 2FA in a widely used admin tool, plus state actors automating vulnerability research at scale.
May 07, 2026
On 7 May 2026 the Council and Parliament agreed the Digital Omnibus on AI, pushing stand-alone high-risk duties to December 2027 and embedded-product duties to August 2028.
Two critical Semantic Kernel bugs let a crafted prompt execute code or write arbitrary files on the agent host. Microsoft says the pattern recurs across agent frameworks.
