Anthropic CEO Dario Amodei warns of a critical 6-12 month window to patch tens of thousands of AI-discovered software vulnerabilities before China's AI capabilities catch up, as the company's restricted Mythos model has already uncovered nearly 300 flaws in Firefox alone.
Apple is set to revolutionize iPhone AI in iOS 27, allowing third-party models from Google, Anthropic, and OpenAI to power Siri and core features system-wide, with users able to customize their preferred AI engine and even assign unique Siri voices to different models.
OpenAI rolls out GPT-5.5 Instant as ChatGPT's new default model, boasting 52% fewer hallucinations on high-stakes topics like medicine and law, more concise responses, and new Gmail integration that personalizes replies using past conversations.
Microsoft ends its exclusive deal with OpenAI as AWS immediately launches OpenAI models, Google Cloud breaks $20 billion in quarterly revenue, Elon Musk admits in court that xAI trained Grok using OpenAI models, and Meta's business AI explodes from 1 million to 10 million weekly conversations in under three months.
Anthropic and OpenAI are launching rival enterprise AI ventures worth billions, with Anthropic securing a $1.5 billion joint venture backed by Blackstone and Goldman Sachs, while OpenAI raises $4 billion for its $10 billion-valued Development Company, as both AI giants race toward potential IPOs.
OpenAI's GPT-5.5 launches with doubled token prices over GPT-5.4, driving real-world cost increases of up to 92% for users, though longer prompts see partial relief as the model generates significantly fewer completion tokens.
OpenAI overhauled its WebRTC architecture with a split relay-plus-transceiver model, enabling real-time voice AI to scale globally across Kubernetes for over 900 million weekly active users while minimizing latency without requiring custom client modifications.
Voice-Pro goes fully open-source, giving everyone free access to powerful AI voice cloning, speech recognition, multilingual text-to-speech, and vocal isolation tools as its developers shift focus to a new global communication platform.
A new open-source framework called open-slide now lets AI coding agents like Claude Code and Cursor build polished, React-based presentations from simple natural language descriptions, complete with presenter mode, a click-to-comment inspector, and one-command export to HTML or PDF.
Facebook Research unveils Tuna-2, a groundbreaking multimodal AI model that ditches traditional vision encoders in favor of direct pixel patch processing, outperforming predecessors on diverse benchmarks while supporting both image understanding and generation tasks in 7B and 2B parameter sizes.