Anthropic announced Claude Science, an AI workbench for researchers, and said it wants to use Claude to help develop drugs itself. The move puts the company beyond selling AI tools to pharma, with ambitions around neglected-disease research that will still need lab validation, trials, and regulation.
Claude Fable is available again to more users, but early users say the relaunched model feels weaker than the original release. The backlash matters for teams that had treated the model as a high-end reasoning option and now need to check whether quality changed in production workflows.
Anthropic is discussing a custom AI chip project with Samsung, according to TechCrunch. A deal would extend the frontier-lab push into dedicated silicon and follows OpenAI's recent custom-chip partnership with Broadcom, underscoring how model providers are trying to control more of their compute stack.
PTA confirmed the SEA-ME-WE 5 international submarine cable is fully restored, returning Pakistan's internet traffic to normal on July 3. The update closes the service degradation that caused slow browsing and intermittent connectivity for users across the country.
Researchers report that Anubis ransomware actors are exploiting Citrix Bleed 2, tracked as CVE-2025-5777, to break into organizations. The campaigns combine credential theft, legitimate remote-management tools, and hands-on lateral movement, so exposed Citrix environments need urgent review and patch verification.
Medtronic is notifying more than 3.8 million people after ShinyHunters accessed corporate IT systems in April. The company says products and manufacturing were not affected, but the stolen personal and medical information makes the incident a high-impact healthcare data breach.
Meta quietly released Pocket, an experimental AI app for generating and sharing interactive mini games from text prompts. The launch shows Meta testing lightweight consumer creation workflows around AI-generated software, not just chatbots and content feeds.
A Reuters report surfaced through Hacker News says Alibaba plans to ban Claude Code internally over alleged backdoor risks. Whether or not the concern proves out, the story highlights the pressure enterprises face when adopting coding agents that can read repositories and execute development tasks.