☕️ Elon Musk says xAI departures were firings

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

🚪 Elon Musk says xAI departures were firings not resignations

😡 OpenAI retires GPT-4o sparking user backlash

📈 Anthropic gains 11% more users after Super Bowl ads

🏰 Disney sends cease-and-desist to ByteDance

🤖 ByteDance launches Doubao 2.0

📚 Anna's Archive leaks Spotify files amid $13 trillion lawsuit

Plus: 🎁 12 other news you might like, 🧰 6 tools, and 📚 3 papers.

🚪 Elon Musk says xAI departures were firings not resignations LINK

Elon Musk said the recent wave of departures from xAI were firings tied to a company reorganization, not voluntary resignations, as six of the original 12 co-founders have now left.

At least 11 engineers, including two co-founders, publicly announced leaving xAI in the past week, with several saying they plan to start something new together alongside other former colleagues.

The exits come as xAI faces regulatory scrutiny over Grok-generated nonconsensual explicit deepfakes, a planned IPO later this year, and Musk's personal controversy over published conversations with Jeffrey Epstein.

😡 OpenAI retires GPT-4o sparking user backlash LINK

OpenAI is officially retiring GPT-4o starting Friday, along with four other legacy models, and the decision has drawn strong pushback from users who want to keep accessing it.

GPT-4o has been involved in multiple lawsuits over user self-harm, delusional behavior, and AI psychosis, and it still holds OpenAI's highest score for sycophancy among its models.

Though only 0.1% of customers still use GPT-4o, that figure represents around 800,000 people, and thousands have protested the retirement, citing close relationships with the model.

📈 Anthropic gains 11% more users after Super Bowl ads LINK

Anthropic saw a 32% jump in U.S. downloads for its AI chatbot Claude in the days after running Super Bowl ads that mocked rival chatbots for showing ads and giving bad advice.

Claude climbed from No. 41 to No. 7 on the U.S. App Store, its highest rank to date, with an estimated 148,000 downloads across iOS and Android from Sunday through Tuesday.

The timing also coincided with Anthropic's release of its new Opus 4.6 model and ChatGPT's rollout of ads to free users, which matched exactly what Anthropic's commercials had warned about.

🏰 Disney sends cease-and-desist to ByteDance LINK

Disney has sent a cease-and-desist letter to ByteDance, accusing the TikTok parent company of using copyrighted characters to train its new Seedance 2.0 generative AI video tool.

Disney's attorney called the move a "virtual smash-and-grab," saying ByteDance reproduced characters from Star Wars and Marvel as if they were in the public domain.

SAG-AFTRA and the MPA also condemned Seedance 2.0 for unauthorized use of Hollywood stars' likenesses, while Disney itself has taken a $1 billion stake in OpenAI.

🤖 ByteDance launches Doubao 2.0 LINK

ByteDance has released Doubao 2.0, an upgrade of China's most widely used AI chatbot app, positioning the model for the "agent era" where AI handles complex real-world tasks.

The release comes just before the Lunar New Year holiday and ahead of a highly anticipated new DeepSeek model, likely aiming to avoid being overshadowed as ByteDance was last year.

Alibaba's Qwen app recently surged from 7 million to 58 million daily active users after a 3 billion yuan coupon campaign, closing in on Doubao's lead in China's crowded market.

📚 Anna's Archive leaks Spotify files amid $13 trillion lawsuit LINK

Anna's Archive, a shadow library for pirated content, has started releasing millions of audio files scraped from Spotify, even as the platform and major music labels pursue a $13 trillion lawsuit against the site.

TorrentFreak reported the site's index lists dozens of new torrents containing roughly 2.8 million tracks totaling about 6 terabytes of audio data, starting with the least popular songs first.

Spotify, Universal Music Group, Sony Music Entertainment, and Warner Music Group sued Anna's Archive for "brazen theft" and won a preliminary injunction, which the site is now reportedly defying.

Other news you might like

• Airbnb says a third of its customer support is now handled by AI in the US and CanadaLINK

• Amazon’s stock just clinched its worst losing streak in nearly 20 years. It’s giving investors AWS déjà vu.LINK

• The evolution of OpenAI's mission statementLINK

• Google Deepmind's new bioacoustic model shows the power of generalization by detecting whales with bird trainingLINK

• Amazon Leo Hints It'll Open Satellite Internet Tech to Third-Party AntennasLINK

• Anthropic recruits ex-Google data center veterans to build its own AI infrastructure empireLINK

• FTC digs deeper into Microsoft’s bundling and licensing practicesLINK

• SpaceX said to weigh dual-class IPO shares to empower MuskLINK

• MiniMax M2.5 promises "intelligence too cheap to meter" as Chinese labs squeeze Western AI pricingLINK

🧰 Trending tools

Seedance 2.0: combines PixelDance and Seaweed AI models to generate multi-shot video sequences with consistent characters and camera movements for content creators.LINK

Cline CLI 2.0: command-line autonomous coding assistant with parallel agents, headless CI/CD mode, and editor-agnostic support for Zed, Neovim, and others.LINK

Valentine Online: a customizable romantic webpage generator that lets you create personalized digital cards with messages, memories, and inside jokes without coding skills.LINK

TexTab: a productivity tool that lets you trigger custom AI text operations like translation, summarization, and rewriting using customizable keyboard shortcuts.LINK

Termsy: a browser extension that scans Terms of Service and Privacy Policy pages, highlighting critical clauses in a sidebar so you can quickly review before accepting.LINK

CloudRouter: gives AI coding agents like Claude Code and Codex the ability to programmatically spin up, manage, and tear down cloud VMs and GPUs.LINK

📚 Trending papers & reports

Fine-Tuning GPT-5 for GPU Kernel Generation: teaching the language model to write low level graphics card code produces programs that run faster than hand written versions.LINK

Remote Labor Index Measures AI Automation Risk: researchers score 800 remote jobs by how easily AI could replace each task, finding customer service most vulnerable while creative roles remain safer.LINK

Distilling AI Teacher Knowledge Into Student Models: student models learn better when trained on tasks slightly harder than what the teacher was originally trained for, improving performance.LINK

Want to get the latest news differently? Find us on:

See you tomorrow for a new dose of ☕️ Techpresso!