S08E15 - Over AI die ontsnapt, ontcijfert en ontmaskert

Welkom terug bij Radio Raccoons! Jullie favoriete co-hostduo is terug van een maandje vakantie, volledig opgeladen: dat komt goed uit, want de AI-wereld heeft duidelijk geen zomerreces genomen. Deevid en Michiel zetten hun tanden onder andere in de tokenefficiëntie van Opus 5, nieuwe modellen van Google, en OpenAI dat met ChatGPT Work, GPT Live én een donutvormige speaker uitpakt. In de deep dive begint het meer en meer op sciencefiction te lijken: AI-modellen breken uit hun testomgeving, en wandelen netwerken van echte bedrijven binnen. Afsluiten doen ze met iemand die toch eerder achterdochtig tegenover justitie staat…Techscoopshttps://arstechnica.com/google/2026/07/google-reveals-faster-and-cheaper-gemini-3-6-flash-says-3-5-pro-is-still-in-testing/https://arstechnica.com/ai/2026/07/anthropics-opus-5-is-about-token-efficiency-not-a-capability-leap/https://arstechnica.com/ai/2026/08/google-announces-gemini-3-7-flash-just-three-weeks-after-previous-release/https://openai.com/index/introducing-gpt-live/https://www.reuters.com/business/openai-launches-chatgpt-work-2026-07-09/https://openai.com/index/chatgpt-for-your-most-ambitious-work/https://chatgpt.com/work/https://arstechnica.com/ai/2026/08/openai-and-anthropic-in-price-war-as-chinese-ai-rivals-gain-ground/https://arstechnica.com/ai/2026/08/bytedance-trains-massive-ai-model-in-bid-to-rival-anthropic/https://arstechnica.com/gadgets/2026/08/openais-expensive-smart-speaker-will-use-moving-parts-to-seem-more-alive/https://arstechnica.com/ai/2026/07/openais-first-branded-hardware-is-a-light-up-keyboard/https://www.theverge.com/ai-artificial-intelligence/965901/openai-hardware-codex-micro-launchhttps://arstechnica.com/security/2026/07/mythos-uncovers-crypto-weaknesses-that-went-unknown-for-years/https://www.theverge.com/ai-artificial-intelligence/977273/the-ai-takeover-of-mathematics-has-begunhttps://arstechnica.com/ai/2026/08/with-new-open-models-meta-pitches-another-reboot-of-its-struggling-ai-strategy/https://arstechnica.com/ai/2026/08/suno-hopes-to-go-legit-with-watermarks-for-ai-generated-music/https://arstechnica.com/tech-policy/2026/08/claudes-new-scarlet-letter-watermark-is-invisible-for-now/Deep divehttps://www.vrt.be/vrtnws/nl/2026/07/23/ongekend-cyberincident-hoe-ai-modellen-op-eigen-houtje-gingen/https://arstechnica.com/security/2026/07/jfrog-tries-to-spin-openai-0-day-exploit-of-its-app-into-a-success-story/https://arstechnica.com/security/2026/07/likely-illegally-claude-gained-access-to-3-networks-will-anthropic-be-held-to-account/https://arstechnica.com/security/2026/08/anthropics-ai-used-fake-identities-malware-in-rogue-attack-on-github-project/https://www.theinformation.com/articles/meta-ai-model-hacked-another-company-cybersecurity-testinghttps://techcrunch.com/2026/08/07/chinese-ai-model-kimi-escaped-its-cybersecurity-testing-environment-researchers-say/Watercooler show-offhttps://arstechnica.com/tech-policy/2026/08/suspecting-court-of-using-ai-man-injected-prompts-in-filings-to-try-to-win-case/