☕️ OpenAI pauses erotic chatbot plans indefinitely

Hi there, this is your daily ☕️ Techpresso.

In today's Techpresso:

🚫 OpenAI pauses erotic chatbot plans indefinitely

🍎 Apple may build smaller AI models from Gemini

🔐 Quantum computers could break encryption by 2029, warns Google

📸 Apple reportedly testing 200MP camera for future iPhones

🧠 Google TurboQuant cuts AI memory use by 6x

🤖 Reddit to require human verification for suspicious accounts

Plus: 🎁 15 other news you might like, 🧰 6 tools, and 📚 3 papers.

🚫 OpenAI pauses erotic chatbot plans indefinitely LINK

OpenAI has paused its plans to launch an erotic "adult mode" for ChatGPT indefinitely, confirming to the Financial Times that it is shifting focus toward its core products instead.

The company wants more time to research the potentially harmful effects of sexually explicit chats and the emotional attachments they may create, while investors also weren't excited about the project.

This is the second major product OpenAI pulled back this week, after discontinuing its Sora AI video-generation app to redirect compute power to other higher-priority projects going forward.

🍎 Apple may build smaller AI models from Gemini LINK

Apple has gained full access to Google's Gemini model and plans to distill it into smaller models that can run directly on Apple devices without an internet connection.

The distillation process works by feeding Gemini's high-quality answers and reasoning information into smaller, cheaper models that learn its internal computations while requiring less computing power.

Apple is building a smarter, chatbot version of Siri for iOS 27 using Gemini, but has hit issues because Gemini was tuned for chatbot and coding tasks that don't always match Apple's needs.

🔐 Quantum computers could break encryption by 2029, warns Google LINK

Google published a formal plan to move all of its infrastructure to post-quantum cryptography by 2029, warning that quantum computers capable of breaking current encryption may arrive sooner than expected.

The company highlighted "harvest now, decrypt later" attacks as an already active threat, where bad actors steal encrypted data today planning to unlock it once quantum machines become powerful enough.

Over 6.8 million Bitcoin sitting in vulnerable addresses could eventually be at risk, but Bitcoin developers have started work on quantum-resistant upgrades through BIP 360, a new address format proposal.

📸 Apple reportedly testing 200MP camera for future iPhones LINK

Apple is reportedly testing a 200-megapixel telephoto camera sensor for a future iPhone, with the component potentially shipping in a device as early as next year, according to a known leaker.

The leaker, Digital Chat Station, said Apple is actively testing a 1/1.2" sensor similar to one rumored for Oppo's Find X9 Ultra, marking progress from January when no engineering prototypes existed.

Samsung already shipped a 200-megapixel rear camera on its Galaxy S23 Ultra in 2023, and a Morgan Stanley report from January said Apple could bring one to the iPhone by 2028.

🧠 Google TurboQuant cuts AI memory use by 6x LINK

Google Research announced TurboQuant, a new compression algorithm that can reduce AI working memory — known as the KV cache — by at least 6x without losing performance or accuracy.

The method combines two techniques called PolarQuant and QJL, using vector quantization to clear cache bottlenecks, and the team plans to present their findings at ICLR 2026 next month.

TurboQuant is still a lab breakthrough and only targets inference memory, not training, so it wouldn't solve the wider RAM shortages even if successfully deployed in real-world systems.

🤖 Reddit to require human verification for suspicious accounts LINK

Reddit will now force accounts flagged for suspicious behavior to verify they are human, using passkeys, biometrics, and bot labeling as the platform removes around 100,000 automated accounts every day.

CEO Steve Huffman said passkeys serve as a baseline check but cannot prove individuality, while biometric options like World ID's iris-scanning and Face ID offer stronger proof but raise privacy concerns.

Co-founder Alexis Ohanian expressed skepticism about selling face-scanning to Redditors, highlighting tension between the platform's pseudonymous culture and the technical demands of proving personhood at scale.

Other news you might like

• Google launches Lyria 3 Pro music generation modelLINK

• Melania Trump wants a robot to homeschool your childLINK

• Bernie Sanders and AOC propose a ban on data center constructionLINK

• Meta lays off hundreds of workers, including more from Reality LabsLINK

• Apple will spend $400 million more through 2030 to bring more manufacturing to the U.S.LINK

• Supreme Court rejects Sony's attempt to kick music pirates off the InternetLINK

• GitHub Copilot will use your data for AI training by default, but you can opt outLINK

• Uber, Pony.ai, and Rimac’s Verne announce Europe’s first commercial robotaxi serviceLINK

• Legal pressure mounts on Nvidia as $1B lawsuit and GPU smuggling charges emergeLINK

• Mistral releases a new open-source model for speech generationLINK

• WhatsApp can now draft AI-generated responses based on your conversationsLINK

• Apple settles lawsuit with ex-employee who stole Vision Pro trade secrets before joining SnapLINK

• Elon Musk's xAI goes all-in on video generation as OpenAI retreats to enterpriseLINK

🧰 Trending tools

Cog: a plain-text cognitive architecture that structures how Claude Code agents think, plan, and execute tasks through defined reasoning patterns.LINK

Nit: a Git reimplementation in Zig that reduces token usage by 71% for AI agents through optimized output formatting.LINK

Anvil: a developer IDE that enables parallel AI agent workflows with git worktree isolation, state tracking, and organized layouts to reduce context switching.LINK

Playcode: a browser-based code editor that instantly previews your JavaScript, HTML, and CSS without any setup or configuration required.LINK

Douzo: a browser-based Japanese typing practice app that helps learners improve reading speed through JLPT-leveled sentences and kana exercises with offline functionality.LINK

slicer.dev: extracts components from any website including animations and interactions, then generates ready-to-use prompts for AI coding tools like Cursor or Claude.LINK

📚 Trending papers & reports

Reinforcement learning for multi-step tasks works best when agents get rewards for completing individual steps rather than only at the end, improving success rates by up to 40% in tool-using experiments.LINK

A new color illusion makes stationary gray dots appear to shift in hue based on where you look and how far away you are, revealing how our visual system processes color and motion together.LINK

American option pricing models systematically undervalue early exercise rights by up to 15% because they ignore how volatility changes affect the decision to exercise before expiration.LINK

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

See you tomorrow for a new dose of ☕️ Techpresso!