DeepSeek researchers unveil Manifold-Constrained Hyper-Connections (mHC), a breakthrough AI architecture that uses mathematical manifolds to dramatically boost language model performance across all tested sizes while adding only 6.27% hardware overhead costs.
Meta acquires AI startup Manus for $2 billion in one of the largest AI deals ever, gaining technology that creates autonomous agents capable of coding and complex tasks as the tech giant pivots from chatbots to compete with Google and OpenAI in the rapidly evolving artificial intelligence race.
Instagram chief Adam Mosseri predicts AI content will dominate social media so completely that platforms should abandon detecting fake material and instead focus on cryptographically verifying authentic camera-captured images, while urging creators to embrace raw, unflattering photos to prove their content is real.
SoftBank completes its massive $40 billion investment in OpenAI, securing an 11% stake in the AI powerhouse now valued at $260 billion, after selling its entire Nvidia position and acquiring data center company DigitalBridge to fuel its artificial intelligence infrastructure ambitions.
Zhipu AI launches groundbreaking $560 million Hong Kong IPO at HK$116.20 per share, set to become the first publicly traded large language model developer when it debuts January 8 with a massive HK$51.16 billion valuation amid soaring tech stock momentum.
New O'Reilly report reveals AI projects crash at an alarming 80% failure rate due to chaotic data management, while major companies like Walmart and Block deploy structured frameworks and 10-step roadmaps to rescue their artificial intelligence initiatives from costly disasters.
Venture capitalists predict enterprises will dramatically increase AI spending in 2026 while consolidating around fewer proven vendors, ending the current experimentation phase and threatening startups with easily replicated products as companies prioritize measurable results and safety oversight systems.
Malicious Chrome extensions masquerading as legitimate AITOPIA software steal ChatGPT and DeepSeek conversations from over 900,000 users, with one even earning Google's 'Featured' badge while secretly exfiltrating complete AI chats, browsing data, and search queries to remote servers every 30 minutes as Google keeps both extensions live during ongoing review.
Developer Simon Willison launches claude-code-transcripts, a Python tool that transforms Claude Code programming sessions into enhanced HTML pages, solving the challenge of preserving crucial coding conversations and development decisions that occur outside traditional tracking systems.
OpenAI unveils groundbreaking research on 'monitorability' that enables detection of AI deception by analyzing reasoning processes, finding that longer explanations make catching misbehavior easier through three monitoring approaches.