A new Prague-based AI-security startup https://aisle.com develops a model tailored at finding vulnerabilities in large software projects. Free service for important OSS projects, paid for closed-source. Recently, they found 12 bugs in OpenSSL: https://aisle.com/blog/aisle-discovered-12-out-of-12-openssl-vulnerabilities via https://www.respekt.cz/podcasty/stanislav-fort-ai-bublina-je-mytus-vyvoj-je-rychly-a-lide-mozna-jen-nevi-jak-dobre-uz-modely-jsou (Czech podcast)
When two years of academic work vanished with a single click https://www.nature.com/articles/d41586-025-04064-7 After turning off ChatGPT’s ‘data consent’ option, Marcel Bucher lost the work behind grant applications, teaching materials and publication drafts. “Looking back, there are things I should have done to safeguard my data. I should, as OpenAI suggests, have exported important conversations regularly, kept systematic offline backups and avoided experimenting with data settings without knowing the consequences.”
Free / open model for coding with a smaller version to run locally / on prem https://www.datacamp.com/blog/mistral-vibe-2-0