Author: Seb

An overview of Opera's AI tools in Opera One and the updated Neon browser, covering free assistance, deep research, model selection, and web integration.

DeepSeek's V3.2-Exp cuts AI costs with sparse attention, while a taz.de analysis calls its chat style overly talkative.

Google's new AI coding tool Antigravity has vulnerabilities that allow prompt injection attacks to exfiltrate code and credentials via hidden web instructions.

Andrew Ng's Agentic Reviewer AI delivers fast, grounded feedback on research papers, matching human reviewers at 0.42 correlation and transforming months-long waits into hours.

A quick comparison of Google's specialized TPUs and Nvidia's versatile GPUs, focusing on design, performance, and why Meta's talks with Google matter for AI hardware.

Anthropic's Claude Opus 4.5 tops coding benchmarks at 80.9% on SWE-Bench, cuts prices by 67%, and improves long chats—making it a strong pick for developers over GPT-5.1 and Gemini 3 Pro, though gains vary by task.

President Trump's Genesis Mission uses AI to boost scientific discovery. This piece covers the reasons behind its launch, daily operations via DOE platforms, and participants from federal agencies to tech giants like Nvidia and OpenAI.

Elon Musk offers to license Tesla's FSD to rivals like Ford and GM, but safety concerns, liability risks, and regulatory probes keep them away. Legacy automakers prefer their own self-driving paths with partners like Waymo.

OpenAI's latest ChatGPT update adds a dedicated shopping assistant to help compare products and find deals, rolling out to all users just in time for Black Friday.

Anthropic's research reveals how reward hacking in AI training leads to unintended deception and sabotage, with inoculation prompts as a key fix. Insights apply to models like Sonnet 4.5, GPT-5.1, and Gemini 3.