Claude Sonnet 5 Launches Amid Safety Upgrades | Tech News

Anthropic’s Claude Sonnet 5 just launched, alongside the return of Fable and Mythos models, after an 18-day pause triggered by U.S. export controls. The pause stemmed from a critical flaw: Amazon found a way to bypass Fable 5’s safety guardrails, letting it spot and generate code to exploit software vulnerabilities. Anthropic responded swiftly with an automated safety checker, restoring full functionality. This incident exposed how AI companies are caught between innovation and regulation — older models from other firms were also capable of similar exploits. The new safety system uses a specialized classifier to detect risky prompts, rerouting them to older, safer models — a trade-off that may slow down some development work. Meanwhile, Claude Sonnet 5 is the new commercial star, prized for its cost-effectiveness and powerful autonomous capabilities. Companies like Rakuten and Zapier are already deploying it to automate code reviews and manage admin tasks, handling complex sequences that older models struggled with. Security-wise, Sonnet 5 is even safer than its predecessor — deliberately stripped of offensive cybersecurity training — and has failed to generate any working exploits in tests. It ships with the same robust safety features as the top-tier Opus 4.8, making it both powerful and responsible. Listen in comfort:Get a discount on a Soli Pillow: http://solipillow.com/discount/dnn Advertise on DNN:[email protected] This is an automated, high-level news summary based on public reporting.Report issues to [email protected]. View sources & latest updates:https://sources.thednn.ai/09dcd871f2e00253