Midjourney just teamed up with NYU researchers to tackle AI's biggest creative flaw: bland, repetitive writing. By introducing two new training methods called DDPO and DORPO, they reward rare and unique high-quality responses during fine-tuning on models like Llama 3.1. It turns out the image generator has serious ambitions for text, too.
Meta is testing a "Write with Meta AI" button on Instagram that scans photos to draft three ready-made comment suggestions. Users can tap a pencil icon in the comment bar to pick or refresh replies. While Meta wants to boost interactions, users already worry the tool will replace authentic friend check-ins with low-effort AI slop.
Meta just announced the rollout of Meta AI across Europe, bringing conversational AI in six European languages to WhatsApp, Instagram, Facebook, and Messenger for free. Users can summon the assistant in group chats using @MetaAI or simply tap the new blue circle icon to search and brainstorm.
Authors just moved for summary judgment against Meta in a California federal court, alleging the company torrented terabytes of pirated books to train its AI models and uploaded data back to the P2P network while leeching. The plaintiffs mocked Meta's justification as a "Bob Dylan defense"—stealing a little lands you in jail, while stealing a lot makes you king.
A federal judge just allowed an AI copyright lawsuit against Meta to move forward, ruling that authors including Sarah Silverman adequately alleged Meta intentionally stripped copyright management information to hide that Llama was trained on their books. While the judge dismissed a computer fraud claim, Meta still faces key infringement claims.
Meta just kicked off small-scale testing of its first custom AI training silicon, aiming to rein in its gargantuan infrastructure spending and ease its reliance on Nvidia. The MTIA chip is designed to be more power-efficient than general-purpose GPUs, and Meta hopes to deploy it for production training by 2026.
A federal judge just ruled that an AI copyright lawsuit against Meta can proceed, rejecting the company's bid to dismiss the case. Authors allege Meta trained its Llama models on their books and removed copyright information to hide the infringement. The judge tossed a computer fraud claim, but Meta must now defend its fair-use arguments in court.
Meta is reportedly gearing up to launch Llama 4 in just a few weeks, putting a major focus on native voice capabilities. Meta CPO Chris Cox confirmed it will be an "omni" model handling speech and text natively, allowing users to interrupt mid-sentence like ChatGPT's Voice Mode. The rush comes as DeepSeek's open-source gains pushed Meta into overdrive.
Meta researchers just unveiled Avat3r, an AI system that generates an animatable, photorealistic 3D avatar head in minutes using only four phone selfies. It uses a vision transformer to predict 3D Gaussians, drastically lowering the hardware barrier from specialized multi-camera rigs. While rendering currently crawls at 8 FPS on an RTX 3090, it points toward effortless avatar setup for future headsets.
Meta is reportedly planning to launch a dedicated Meta AI mobile app during the second quarter, taking direct aim at OpenAI's ChatGPT. The company also intends to test a paid subscription tier. While Meta AI already reaches hundreds of millions inside Facebook and WhatsApp, a standalone app could unlock deeper personalization and hardware sync.
Meta just unveiled Aria Gen 2, its next-generation smart glasses built for machine perception and robotics research. The 75-gram frames run SLAM and multi-modal tracking entirely on-device via custom silicon, packing new nosepad sensors to isolate the wearer's voice and pulse. Why it matters: It provides AI researchers with a wearable testbed to collect egocentric data for training embodied AI and assistive tech.
Meta just announced LlamaCon, its first developer conference dedicated entirely to generative AI, set for April 29. The event will showcase the latest in its open-source AI roadmap as Mark Zuckerberg prepares to launch reasoning and multimodal Llama models to fend off rising open-weight competition like DeepSeek.
Meta researchers just published two papers showing how AI can decode brain activity into text using non-invasive MEG and EEG sensors. Its deep learning model, brain2qwerty, reached up to 80% character accuracy across 35 typing subjects. It brings surgery-free mind-to-text communication and speech restoration a big step closer to reality.
Meta is making a major bet on humanoid robots, setting up a dedicated team inside Reality Labs led by former Cruise CEO Marc Whitten. Instead of rushing to sell its own branded bot, Meta wants to build the underlying AI, sensor, and software platform using Llama to power third-party hardware, with talks already underway with Figure AI and Unitree.
Arm is reportedly preparing to launch its first in-house server CPU as soon as this summer, with Meta lined up as an early customer. Driven by SoftBank's Masayoshi Son, the move marks a dramatic shift away from Arm's pure-play licensing model and pits it squarely against Intel and AMD in the AI data center race.
Meta is reportedly in talks to acquire South Korean chip startup FuriosaAI, with an announcement possible as soon as this month. FuriosaAI builds chips that speed up model serving, including text models like Llama 3. The move highlights Meta's push to cut reliance on Nvidia as its AI budget climbs to $65 billion this year.
MIT and Whitehead Institute researchers just introduced ProtGPS, a deep-learning model that predicts how proteins sort themselves within cellular condensates. Built on Meta's ESM framework, it complements AlphaFold's structural predictions by decoding spatial organization rules, paving the way for targeted drug discovery.
Meta is preparing a major wearables push for 2025, working on sports-focused Oakley smart glasses and a high-end pair with a built-in lens display dubbed Hypernova. The company is also exploring camera-equipped earbuds and smartwatches, aiming to weave its AI into more everyday hardware.
OpenAI and other leading AI labs are preparing to unveil Ph.D.-level "super-agents" capable of independently tackling messy, multilayered human jobs, according to Axios. OpenAI CEO Sam Altman has even scheduled a closed-door briefing with U.S. government officials on Jan. 30. Why it matters: If these autonomous agents can overcome nagging hallucination issues, generative AI moves from an assistant to a real replacement for mid-level knowledge workers.
Mark Zuckerberg used YouTube's fight against piracy to justify Meta's use of copyrighted datasets during a deposition in an ongoing AI copyright lawsuit. Newly unsealed court filings allege Meta trained its Llama models on pirated e-book repositories like LibGen and Z-Library despite internal legal warnings.
