4 hours ago
ChatGPT Can Now Read, Write, and Send Your iMessages, and Apple Isn’t Happy About It
Key Takeaways OpenAI launched an Apple Messages plugin that lets ChatGPT read, search, draft, and send iMessages on Mac. The plugin requires deep macOS permissions, including Full Disk Access, Contacts access, and automation tools. OpenAI says the plugin runs locally, but message data still reaches ChatGPT when generating summaries or…
23 hours ago
Australia Passes Law Forcing Meta, Google, TikTok and LinkedIn to Pay for News or Face 2.5% Levy
Key Takeaways Australia’s parliament passed the News Bargaining Incentive on Thursday, targeting Meta, Alphabet’s Google, TikTok and Microsoft’s LinkedIn. The levy is set at 2.5% of a platform’s Australian advertising revenue, applied only to firms earning above A$250 million ($178 million) locally. Platforms can offset the charge entirely by striking…
1 day ago
Your Enterprise Data Just Became the Real Battleground Between OpenAI and Anthropic
Key Takeaways OpenAI introduced Private Safety Processing to detect cross-session AI misuse without retaining customer conversations. Anthropic keeps 30 days of data for its most capable models to detect threats across multiple requests. OpenAI is betting that privacy-preserving safety signals can replace the broader data access Anthropic considers necessary. The…
1 day ago
Google Floods Search and Gemini With Free AI Study Tools, Aiming to Beat OpenAI’s Head Start
Key Takeaways Google added AI study tools across Search and Gemini, including interactive visual quizzes and study documents. Eligible US college students can get one free year of Google AI Pro with expanded Gemini features. Google is bringing AI study features directly into Search where students already look for information. …
2 days ago
Cerebras Unveils CS-4, Claiming 30x Nvidia’s Inference Speed as It Races Toward Profitability
Key Takeaways Cerebras launched CS 4, claiming up to 30 times faster chatbot response speeds than Nvidia GPUs. CS 4 uses three wafer-scale chips and 50 percent fewer components than its CS 3 predecessor. Cerebras powers OpenAI Ultrafast mode as it pushes faster AI inference and expands its customer base. …
