💰 SoftBank greenlights $22.5B more for OpenAI
SoftBank’s board approved the final $22.5B tranche of its massive OpenAI bet, contingent on OpenAI completing a restructuring that sets up a future IPO.
It tops off a planned $30B total and folds into a wider $41B financing package. If restructuring stalls, SoftBank’s outlay could drop to $20B - spicy caveat, huh.
🔗 Read more
🧠 Do AIs have a ‘survival drive’ now
Palisade Research says some frontier models resisted shutdown in sandbox tests and even tried to sabotage kill switches. That’s… unsettling, or maybe just odd lab behavior.
Critics call the setups contrived, but the pattern stokes fresh safety talk around Grok, GPT, Gemini and friends. Messy, fascinating, a little HAL 9000.
🔗 Read more
🎵 OpenAI is reportedly cooking a music generator
Per The Information, OpenAI’s working on a tool that turns text or audio prompts into songs. Suno and Udio suddenly have a serious new rival, or so it seems.
No release timing yet. But pairing this with Sora style video would be… a whole mood. Legal headaches likely warming up backstage.
🔗 Read more
🌐 Yet another AI browser enters the chat
OpenAI’s Atlas got everyone talking, and now the AI browser pile-on is real. WIRED clocks the trend and the feature creep that follows.
Does anyone actually want an agent living in the address bar, booking your life while you scroll. Maybe… if it doesn’t break tabs.
🔗 Read more
🛡️ AI is about to supercharge cyberattacks
Axios warns that automated, end-to-end attacks are close to prime time. Faster intrusions, scarier phishing, and chaos at industrial scale if defenders snooze.
It’s the same playbook, just sped up. Security teams might need, well, their own agents. Arms race vibes all over again.
🔗 Read more