Responsible AI · Enterprise engineering · Global delivery

AI News & Blog

Curated analysis of AI, ML, and Agentic developments.

Dec 11, 2025

OpenAI Releases GPT-5.2 With Three Variants and a 30% Cut in Hallucinations

Released on 11 December 2025 after an internal 'code red', GPT-5.2 arrives in Instant, Thinking and Pro variants with 256k context, a $1.75 per million input price and new leads on GDPval and ARC-AGI-2.

#openai#gpt-5#enterprise

Dec 11, 2025

White House Executive Order Targets State AI Laws With Litigation and Funding Levers

Executive Order 14365, signed 11 December 2025, creates a DOJ task force to challenge state AI laws, orders Commerce to list 'onerous' statutes and conditions broadband funds on compliance.

#policy#united-states#regulation

Dec 06, 2025

Researcher Reports 30+ Flaws Across AI Coding Tools, 24 With CVEs (Dec 2025)

A December 2025 disclosure catalogued more than 30 vulnerabilities in Cursor, Copilot, Windsurf, Claude Code, Cline and other AI coding agents, spanning prompt injection, data theft and remote code execution.

#security#agents#supply-chain

Dec 01, 2025

DeepSeek V3.2 Brings Olympiad-Level Reasoning to Open Weights Under MIT Licence

On 1 December 2025 DeepSeek released V3.2 and V3.2-Speciale, 685-billion-parameter MIT-licensed models claiming gold-medal performance on the 2025 IMO. Open weights again narrow the gap with closed frontier systems.

#deepseek#open-weight#research

Nov 26, 2025

OpenAI Discloses Third-Party Breach at Mixpanel Exposing API Customer Metadata

On 26 November 2025 OpenAI told API customers that an intrusion at analytics vendor Mixpanel exposed names, emails, locations and account identifiers. No prompts, keys or payment data were involved, but it is a textbook vendor-risk lesson.

#openai#security#supply-chain

Nov 19, 2025

EU Proposes Digital Omnibus: High-Risk AI Act Obligations Could Slip to 2027

The European Commission's 19 November 2025 Digital Omnibus on AI would tie high-risk obligations to the availability of standards, with backstop dates of December 2027 and August 2028, and widen SME relief to small mid-caps.

#eu-ai-act#policy#regulation

Nov 18, 2025

Google Ships Gemini 3 Across Search, Cloud and a New Agentic IDE (Nov 2025)

Google released Gemini 3 Pro on 18 November 2025, deploying it to Search on day one and launching Antigravity, an agent-first development platform. Benchmarks lead, but agentic risks arrive with it.

#google#gemini#agents

Nov 13, 2025

When AI Runs the Attack: Anthropic’s November 2025 Cyber Espionage Disclosure

In mid-November 2025, Anthropic revealed that its Claude model had been abused in what it calls the first large-scale, largely automated AI-driven cyber espionage campaign—and described how it shut it down.

#anthropic#security#espionage

Nov 12, 2025

Anthropic’s $50 Billion Bet on U.S. AI Infrastructure: What It Means for the Compute Race

On November 12, 2025, Anthropic announced a $50 billion plan to build custom AI data centers in Texas and New York with Fluidstack—another huge escalation in the global race for AI compute.

#anthropic#infrastructure#data-centers