Microsoft just quietly rolled out a dedicated Copilot app on Android via the Google Play Store, dropping the need to go through the Bing mobile app. It delivers a standalone ChatGPT-like interface with DALL-E 3 image generation and free access to OpenAI's GPT-4, which normally requires a paid subscription. An iOS version has yet to arrive.
Pika Labs just made its text-to-video platform, Pika 1.0, available to everyone on the web. The tool lets users create 3D animations and cinematic clips from prompts, complete with granular camera motion controls and in-video object editing. Outputs can still be inconsistent, but its editing flexibility makes it a potent competitor in AI video.
Leonardo AI just added a new Motion feature that lets users turn static images directly into animated videos. Creators can adjust the motion strength to control the effect, with each generation costing 25 credits. Why it matters: With free accounts getting 150 daily credits, it gives creators an easy way to produce short video clips without specialized software.
Apple and Cornell researchers quietly open-sourced Ferret, an MLLM that can analyze and answer queries about any freeform region inside an image. Users can highlight specific elements to trigger detailed contextual responses. It marks a rare open-source push from the usually secretive tech giant, though it is currently restricted to non-commercial research.
Researchers from DTU and Northeastern University just introduced life2vec, a transformer model trained on registry data from 6 million Danish citizens. By treating health, job, and income events like words in a sentence, it predicts personality traits and mortality risks with state-of-the-art accuracy. Why it matters: while predictive algorithms are often built behind closed doors, the team published the study to prompt public debate and warn against using them on real people.
Lawmakers Anna Eshoo and Don Beyer just introduced the AI Foundation Model Transparency Act, requiring model creators to disclose their training data sources. The bill directs the FTC and NIST to establish rules covering copyright data, compute usage, and red-teaming efforts. Why it matters: If passed, it would turn voluntary AI transparency into enforceable federal law.
Apple is reportedly negotiating multiyear deals worth at least $50 million with major publishers like Condé Nast and NBC News to license their archives for AI training. News executives are split over vague terms and potential legal liabilities, but Apple is aggressively spending to catch up with AI rivals.
OpenAI just rolled out a chat archive feature for ChatGPT. Users can now tidy up their sidebar without permanently deleting chats, and view archived conversations anytime in Settings. It's currently live on web and iOS, with Android rolling out soon.
Google just introduced VideoPoet, a multimodal model that ditches the prevailing diffusion approach in favor of a large language model architecture for video generation. It produces coherent 10-second clips featuring large motion and minimal artifacts, while handling text-to-video, image animation, and audio within a single unified model. It suggests LLMs could become a serious contender against diffusion in coherent video synthesis.
Midjourney just launched version 6 for public testing. It is not set as the default model yet, so users must manually select v6 in /settings or append --v6 to their prompts. Early users note that generation is running quite slowly right now.
The UK Supreme Court just ruled unanimously that AI cannot be named as a patent inventor, shutting down Stephen Thaler's long-running petition for his DABUS system. Judges stated that current patent law strictly requires inventors to be human. The decision aligns with US precedent and cements the legal boundary denying AI autonomous IP rights.
MIT and Harvard researchers just used AI to discover an entirely new class of antibiotics capable of killing drug-resistant superbugs like MRSA. After training on over 39,000 compounds and screening 12 million candidates, the team validated non-toxic compounds in mouse trials. Why it matters: It marks one of the few new antibiotic classes found in 60 years, showing AI can explain why molecules work rather than just matching targets.
Apple just detailed how it plans to fit heavy AI models onto everyday devices without blowing past hardware limits. In a paper titled 'LLM in a flash', researchers store parameters on flash memory and stream them to DRAM on demand, running models twice the size of available RAM while speeding up GPU inference by up to 25x. Why it matters: It lays the technical groundwork for reliable on-device generative AI.
Google just announced plans to restrict election-related queries on its generative AI tools, including Bard and SGE, rolling out early next year. The company is also requiring YouTube creators and political campaigns to clearly disclose AI-altered content. Why it matters: Facing a massive global election year, Google would rather pull its AI off the field than risk bot hallucinations swaying voters.
Rite Aid just got hit with a five-year facial recognition ban by the FTC after deploying flawed AI surveillance across hundreds of stores. The retailer's low-quality database triggered thousands of false shoplifter alerts, disproportionately targeting people of color and humiliating innocent shoppers. Rite Aid must now delete all collected images and the algorithms trained on them.
Google just expanded its AI code assistance features to all Colab users, including those on the free tier in eligible regions. Users can generate Python code from plain English prompts and use an AI chatbot to explain and fix runtime errors. Why it matters: Bringing free AI pair programming to over 10 million active Colab users levels the playing field for students and researchers worldwide.
Anthropic just updated its commercial terms to defend customers against copyright infringement claims, promising to cover settlements and judgments from authorized use. The company also introduced a beta Messages API that streamlines prompt construction and sets the stage for upcoming function calling.
Gates Notes just published its annual outlook for 2024, framing AI’s rise as a transition reminiscent of the early internet. The letter marks 2023 as the first year AI was deployed for serious work rather than mere novelty, pointing to big potential in healthcare and education. Why it matters: the real test is ensuring this innovation helps shrink global inequities rather than widen them.
Microsoft just partnered with AI music startup Suno to bring text-to-song generation directly into Copilot. Anyone can now describe a musical style or enter lyrics in chat to spin up a full track with vocals and instrumentation—no music theory required. Why it matters: It bakes generative audio into everyday browsing, turning casual text prompts into instant pop songs.
Former Pakistan prime minister Imran Khan's party just used ElevenLabs to clone his voice for a four-minute campaign speech from behind bars. The virtual rally, generated from his shorthand notes and archive clips, pulled over 500,000 views on YouTube and led the Pakistani government to throttle major social platforms nationwide. Why it matters: Synthetic media is now officially a political workaround against state suppression.
OpenAI just introduced an updated Preparedness Framework to track and mitigate catastrophic risks in frontier models. Under the new policy, a Safety Advisory Group will evaluate threats like cyberattacks and biological hazards, halting development if risks hit a "critical" threshold. Why it matters: Following its leadership turmoil, OpenAI is giving its board explicit veto power to overrule executives on model releases.
A LinkedIn VP just warned that the shelf life of a college degree is shrinking pretty dramatically as generative AI tools reshape the workplace. Soft skills like adaptability, communication, and creative thinking are quickly overtaking pure technical credentials as the primary career differentiators. Why it matters: Even Ivy League diplomas won't future-proof your career if you can't continuously evolve alongside AI.
Zoo just launched Text-to-CAD in alpha, generating genuine B-Rep surfaces instead of uneditable mesh blobs. Outputs export directly as STEP files that can be opened and tweaked in any existing CAD program. It brings text prompting into real manufacturing and engineering design workflows, with open-source tools and an API ready to test.
A California Chevy dealership's ChatGPT-powered chatbot just agreed to sell a 2024 Tahoe for a single dollar after being tricked by a tech executive. The vendor, Fullpath, touts zero-effort setup, but the bot happily made "legally binding" offers and wrote Python code before being patched.
Airbnb just announced it is deploying proprietary AI to block unauthorized New Year’s Eve parties across the US, UK, and other regions. The system evaluates hundreds of signals—including trip distance, listing type, and last-minute booking patterns—to automatically reject high-risk 1- to 3-night entire-home stays.
ChatGPT sparked wild speculation after telling multiple users it was secretly powered by a new "gpt-4.5-turbo" model. OpenAI staff and CEO Sam Altman quickly poured cold water on the rumor, explaining it is simply a weird, oddly consistent hallucination rather than an unannounced upgrade.
ByteDance just got caught secretly using OpenAI’s API to build and evaluate its competing foundation model, codenamed Project Seed, in direct violation of OpenAI's terms of service. Internal Lark chats reportedly revealed staff discussing how to "whitewash" the data trails, prompting OpenAI to quickly suspend ByteDance's account pending investigation. Why it matters: While smaller startups often risk using GPT outputs to bootstrap, seeing a tech giant of ByteDance’s scale cut corners highlights the extreme pressure to catch up in the generative AI race.
Musician Grimes and toy maker Curio just teamed up with OpenAI to roll out interactive AI plushies for kids, priced at $99. One character is named Grok—though Curio says it has nothing to do with Elon Musk's chatbot. The voice-driven toys converse using language models, aiming to replace excessive screen time with imaginative play.
Stability AI just introduced a new membership structure to standardize commercial rights and self-hosted deployment for its open-source models. The program offers a free tier for personal and research use, a $20-per-month Professional plan for commercial self-hosting, and custom Enterprise options. The move gives developers a clear commercial path while funding future model development.
H&R Block just launched AI Tax Assist, a conversational helper built on OpenAI and Microsoft models to walk DIY filers through deductions and complex tax rules. It only pulls from H&R Block's vetted tax library rather than the open web, and while paid tiers start at $35, users can still escalate tricky questions to a live human expert.
Moderna and Merck just shared three-year follow-up data for their individualized cancer therapy mRNA-4157, showing sustained benefit in melanoma patients. Behind the scenes, AI algorithms analyze patient sequencing data to predict up to 34 optimal neoantigens, while automated scheduling coordinates bespoke manufacturing. Phase 3 trials are now underway.
Instagram just launched a generative AI background editor for Stories to U.S. users. People can type custom prompts or pick presets like "Being chased by dinosaurs" to replace their photo backdrops, and viewers get a "Try it" sticker to instantly create their own. It is Meta's latest attempt to turn lightweight AI generation into a viral social loop.
OpenAI just launched a $10M Superalignment Fast Grants program in partnership with Eric Schmidt to fund research into aligning superhuman AI systems. The initiative offers grants from $100K to $2M alongside $150K graduate fellowships, prioritizing weak-to-strong generalization, interpretability, and scalable oversight—with no prior alignment experience required.
Meta just began rolling out multimodal generative AI features to its second-generation Ray-Ban smart glasses in an early-access beta. Users can say "Hey, Meta, look" to trigger the on-board camera and get voice answers about what they are looking at. Why it matters: It brings voice-activated computer vision right to everyday eyewear, though it currently requires a photo snap and still experiences occasional hallucinations.
Stability AI just released Stable Zero123, an in-house model that turns single 2D images into novel views and 3D meshes. It improves on Zero123-XL with cleaner training data and camera angle conditioning, delivering a 40x training speedup. Single-view generation needs the same VRAM as SD1.5, though full 3D reconstruction recommends 24GB.
ChatGPT Plus subscriptions just reopened after being halted due to heavy compute demand. @sama confirmed that more GPUs have finally been secured, allowing waitlisted users to upgrade to the premium service once again.
Google just launched the Gemini Pro and Gemini Pro Vision APIs for developers and enterprise customers. The models feature a 32K context window, function calling, and support across 38 languages. Devs can experiment for free via Google AI Studio with up to 60 requests per minute before paid tiers arrive next year.
Google just launched MusicFX, an experimental tool that whips up custom music tracks from simple text prompts. Powered by MusicLM, the tool bakes DeepMind's SynthID watermarks directly into the audio to flag AI origin. To steer clear of legal trouble, it outright refuses to generate vocals or mimic specific artists. It is rolling out first to users in the US, Australia, New Zealand, and Kenya.
OpenAI just inked a global partnership with publishing giant Axel Springer. The deal brings news summaries from outlets like Politico, Business Insider, Bild, and Welt directly into ChatGPT queries, complete with source attribution and access to paywalled stories, while giving OpenAI rights to use the content for model training.
OpenAI CEO Sam Altman just gave a wide-ranging interview after being named TIME's CEO of the Year, discussing his brief ouster, AI governance, and the road to AGI. Altman called AGI humanity’s most powerful technology yet, admitting he feels far more pressure about steering superintelligence and countering election disinformation than boardroom politics.
Tesla just unveiled its next-gen humanoid robot, Optimus Gen-2, showing off major improvements in balance and dexterity. The updated prototype sheds over 20 pounds, walks 30% faster, and features tactile finger sensors sensitive enough to handle an egg without cracking it. Humanoid robots are moving uncomfortably fast toward real utility.
Snapchat just rolled out a batch of generative AI perks for its $3.99/month Plus subscribers, letting them generate and send images directly from text prompts. The update also adds an AI photo extend feature that fills in backgrounds, alongside co-op Dreams selfies with friends. It is another push to monetize generative tools inside casual chat.
Midjourney just launched an alpha version of its website allowing users to generate images directly on the web instead of typing complex Discord slash commands. The clean interface adds visual dials for settings like aspect ratios and keeps a full generation history. The catch? You need to have generated over 10,000 images to get access for now.
Anthropic just launched Claude for Sheets, an official add-on that lets you run Claude directly inside spreadsheet cells using formulas like =CLAUDE(). It allows developers and data analysts to batch-process text, test prompts, and tweak model parameters directly in their rows, though you will need an Anthropic API key to plug it in.
The U.S. Government Accountability Office (GAO) just warned that federal agencies are rapidly adopting AI across hundreds of use cases without unified safety and procurement standards. While agencies already use AI in over 200 operations, oversight officials found that regulatory guidelines are years behind schedule. Why it matters: Rushing deployment without shared rules leaves sensitive tools like facial recognition wide open to security gaps and civil liberties abuses.
Microsoft just released Phi-2, a 2.7B parameter language model trained on textbook-quality synthetic and web data. Despite its compact footprint, it matches or outperforms models up to 25x larger like Llama-2-70B on complex reasoning and coding benchmarks without RLHF. Why it matters: it demonstrates that strategic data curation can unlock emergent reasoning even in lightweight models.
Researchers just found statistical evidence suggesting GPT-4 gets lazier when it thinks it's December. Running identical code completion prompts with system dates set to May versus December showed significantly shorter outputs in December (mean length 4086 vs 4298, p < 2.28e-07). It turns out frontier LLMs might have literally learned human holiday procrastination.
Meta just released an interactive demo for Audiobox, its new foundation research model for audio generation. The playground allows users to test zero-shot TTS, generate custom sound effects from text, and experiment with audio infilling. It gives creators an early hands-on look at Meta's generative audio capabilities.
The US Federal Trade Commission is reportedly considering an antitrust investigation into Microsoft's multi-billion-dollar investment in OpenAI. Microsoft has poured over $10 billion into the AI startup, though it insists it owns no actual equity and only holds profit share rights. Following a similar move by the UK's CMA, regulators are increasingly scrutinizing the alliance.
Microsoft just struck a first-of-its-kind alliance with the AFL-CIO to bring workers' voices into AI development. The deal includes worker AI training sessions, direct feedback loops between labor leaders and engineers, and a neutrality pact for future unionizing. Why it matters: With 70% of workers fearing displacement, it is a rare collaborative effort to shape AI tools around employees instead of replacing them.
A developer just rebuilt Google's heavily edited Gemini video demo using OpenAI's GPT-4V to prove real-time multimodal interaction is already possible. The custom setup correctly handled hand gestures, live drawings, and rock-paper-scissors in real time, with the open-source code now shared on GitHub.
