Google just fired back against reports claiming it scraped ChatGPT data to train its Bard chatbot. A report from The Information alleged an engineer quit for OpenAI after warning against using ShareGPT data, but Google firmly denies Bard used any rival data—though it dodged whether it ever tried in the past.
Elon Musk, Steve Wozniak, and top AI researcher Yoshua Bengio just signed an open letter calling on all AI labs to immediately halt training systems beyond GPT-4 for at least six months. The letter warns of an out-of-control race with profound societal risks, urging regulators to step in if labs refuse to pause.
Microsoft just launched Microsoft Security Copilot, a cybersecurity assistant combining OpenAI's GPT-4 with proprietary security models trained on 65 trillion daily threat signals. Defenders can run natural-language queries to investigate incidents, summarize alerts, and reverse-engineer threats in minutes instead of days. Why it matters: with a global shortfall of 3.4 million cyber pros, it gives overwhelmed defenders an AI-scale safety net.
Zoom just partnered with OpenAI to supercharge its Zoom IQ assistant, letting latecomers ask AI to summarize what they missed and answer follow-up questions. It can also draft chat responses and create whiteboards from text prompts. It is Zoom's direct answer to Microsoft Copilot, though email and chat drafts roll out invitation-only in April.
OpenAI just published a technical post-mortem detailing the March 20 ChatGPT outage. A bug in the open-source redis-py library corrupted cache connections, accidentally exposing chat titles and payment info—including names, emails, and credit card last four digits—for about 1.2% of active Plus subscribers. OpenAI has patched the issue alongside Redis maintainers and restored service.
OpenAI just launched plugins for ChatGPT, allowing the AI to browse the web, run Python code, and connect with third-party services like Expedia and Zapier. Acting as "eyes and ears" for the model, plugins solve knowledge cutoff and hallucination limits. Access is currently rolling out to waitlisted developers and ChatGPT Plus subscribers.
Canva just debuted a heavy slate of AI features at Canva Create, led by OpenAI-powered Magic Design to spin up curated templates and presentation decks from a single prompt. The refresh also adds Magic Edit for instant object swaps, expands Magic Write across all formats, and bumps text-to-image resolution by 16x.
GitHub just unveiled Copilot X, a major vision powered by OpenAI's GPT-4 that moves AI far beyond basic code completion. It integrates a ChatGPT-style chat directly into IDEs, automatically drafts pull request descriptions, answers questions on documentation, and assists in the CLI. Basically, your AI pair programmer is stepping up to cover the full software development lifecycle.
Microsoft just integrated Bing Image Creator into the new Bing preview and Edge browser, powered by an advanced DALL∙E model from OpenAI. Users in Creative mode can now generate custom pictures right inside search chat simply by describing what they want. It basically transforms web search from tracking down existing images into conjuring new ones from scratch.
Bill Gates just published a memo calling modern artificial intelligence the most revolutionary technology he has seen since the graphical user interface in 1980. After watching OpenAI's GPT ace an AP biology exam with a top score of 5, Gates argued that AI will reshape healthcare and education while helping eliminate global inequities.
OpenAI CEO Sam Altman and CTO Mira Murati sat down with ABC News to discuss GPT-4 and the societal shifts ahead. While calling AI humanity's greatest technological leap, Altman admitted he is genuinely worried about large-scale disinformation and automated cyberattacks. Altman stressed that society has limited time to figure out safety and regulation before less cautious players enter the race.
LinkedIn just rolled out AI writing assistants to help job seekers spruce up their profiles and recruiters draft job descriptions. Profile summaries run on GPT-4 for Premium subscribers, while job posts are powered by GPT-3.5. Why it matters: LinkedIn won't flag AI-generated text, making the platform feel a lot more automated.
The PyTorch team just rolled out PyTorch 2.0, introducing a major compiler revamp that boosts performance without breaking backward compatibility. The marquee feature, torch.compile, leans on OpenAI Triton to deliver an average 36% speedup under AMP precision across 165 models. Native scaled dot product attention is also baked in, eliminating the need for external FlashAttention packages.
Stripe just teamed up with OpenAI to power global checkout, recurring billing, and tax compliance for ChatGPT Plus and DALL·E. In return, Stripe is integrating OpenAI’s newly launched GPT-4 into its own platform, kicking off with an AI-powered documentation search that lets developers query technical guides in plain English.
OpenAI just launched GPT-4, a multimodal model that accepts text and image inputs while showing human-level benchmark performance. It scored in the top 10% of a simulated bar exam—trouncing GPT-3.5's bottom 10% showing—and cuts disallowed responses by 82%. Text capabilities are live for ChatGPT Plus and API waitlists, while image features remain in research preview.
Duolingo just announced Duolingo Max, a new subscription tier powered by OpenAI’s newly unveiled GPT-4. The upgrade introduces two conversational AI features: Explain My Answer, which breaks down mistakes in a chat format, and Roleplay, letting learners practice real-world scenarios like ordering at a cafe. It is rolling out now on iOS in the US and select markets.
Microsoft just confirmed that its new Bing search engine has been running on OpenAI's newly released GPT-4 all along, customized specifically for search. Anyone who used the Bing preview over the past five weeks was already testing an early version, with future model upgrades set to flow directly into the search engine.
Microsoft just laid off its entire AI ethics and society team as part of its recent 10,000-person job cuts. The group was tasked with ensuring ethical principles made it into actual product design, including flagging risks around rapid OpenAI integrations. Why it matters: As leadership pressures teams to ship AI at breakneck speed, the internal guardrails are being dismantled, even while Microsoft claims its policy office remains intact.
Local LLMs just became real on everyday hardware. Thanks to Georgi Gerganov’s new llama.cpp and 4-bit quantization, Meta’s LLaMA can now run directly on personal laptops like a MacBook. It brings GPT-3-class capabilities straight to individual developers, breaking AI out of the corporate cloud garden.
General Motors is developing a virtual in-car assistant powered by OpenAI tech via Microsoft's Azure cloud. Moving beyond basic voice controls, the assistant is designed to diagnose dashboard warning lights, play step-by-step videos for changing a flat tire, and schedule repair appointments directly from your vehicle.
