Meta reportedly built a chore robot

Hi there, this is your daily ☕️ Techpresso.

Meta built a robot to do your chores LINK

• Meta has built a prototype humanoid robot, codenamed Max, designed to handle household chores, created in about three months with help from the Chinese robotics firm Fourier, according to The Information.
• Max weighs around 48kg, stands roughly 1.5m tall, and runs on NVIDIA's Jetson Thor chips, with a display for facial expressions, a speaker, and hands made to pick up small objects.
• Meta is already building a follow-up called Max 2, in the works since September, with a redesigned torso and hands plus a projector and extra sensors to help it climb stairs.

Musk admits Grok Bot costs too much to run LINK

• Elon Musk has acknowledged that his company's AI assistant, Grok Bot, costs too much to run compared to the money it makes, and he declined to raise the usage limits that X users have been hitting.
• Running Grok Bot requires heavy computing hardware, data centers, and electricity, and those bills climb as more people sign up and send harder requests to the service, which now has about 800,000 paying users.
• Musk made no firm promises, but his comments suggest xAI, the company behind Grok Bot, may change its pricing, usage caps, or setup, leaving heavy users facing restrictions until a workable model is found.

OpenAI merges ChatGPT and Codex usage limits LINK

• Starting in December 2026, OpenAI will end separate usage limits for Plus, Pro, and Business subscribers, merging ChatGPT and Codex into one shared allowance, though Instant chat will stay outside the combined pool.
• The shared allowance will be counted in tokens, the units a model reads and writes, so heavier models and tougher tasks eat more of it, with subscribers able to check remaining usage and reset times under Settings.
• Users who hit the limit can wait for a reset, upgrade their plan, or buy credits, a shift following October 29 changes that doubled the Work and Codex allowance on Pro 100 and Pro 500 plans.

Hackers heat supermarket freezers to 50°C across Europe LINK

• Hackers broke into the fridge and freezer systems at dozens of European supermarkets last night, remotely raising temperatures to 50°C (122°F) and ruining the food inside at stores across at least seven countries.
• Around 40 Delhaize shops in Belgium and Luxembourg were hit, along with ICA, Coop, SuperValu, Carrefour, and Spar stores in Sweden, Ireland, the Netherlands, Estonia, and Finland, with one Belgian store binning €100,000 worth of food.
• The attack echoes one last month on roughly 400 French supermarkets targeting refrigeration software from Danish firm Danfoss, which said the hacked machines were too old to patch, as experts urge stronger security and regular updates.

Apple pulls 19 apps making deepfake nudes of victims LINK

• Apple has pulled 19 apps from its App Store that hid AI deepfake nudity tools behind innocent features like pet portraits, after a WIRED investigation exposed a network built to slip past the company's review process.
• The apps contacted remote servers after installation, sending each user's language and referral source so the server could decide whether to show advertised features or load hidden websites generating explicit images, often triggered by Facebook or Instagram ads.
• Researchers linked more than 400 apps and domains to the operation, tied some to a Hong Kong firm called KeLing Technology, and found 27 related apps on Google Play with over 350,000 downloads combined.

Anthropic stops Claude from cloning specific games LINK

• Anthropic has added new rules to Claude Code that stop its AI from recreating specific existing games, even when they are renamed or passed off as original projects.
• Claude can still build original games, work within a genre, and use a person's own designs and assets, according to Anthropic's Addy Osmani, who confirmed the change after it was first seen on X.
• The restrictions follow a wave of games like Halo: Combat Evolved, GTA: Vice City, and Call of Duty: Black Ops being rebuilt for web browsers with AI, pushing indie studios such as Digital Sun to pull public Steam demos.

Other news you might like

• SpaceX reportedly in talks to buy EchoStar’s Boost Mobile LINK

• AI agent swarms unleashed in Minecraft to ‘build a thriving civilization’ — 50-agent teams fail after falling into paralyzing bureaucracy LINK

• Waymo picks Toyota bZ4X as its next robotaxi, starting in Japan in 2027 LINK

• Behind the Curtain: OpenAI's post-election power play LINK

• Saudi Arabia turns to Aramco to fuel AI ambitions LINK

• Sam Altman meets Samsung’s presidents days after his DRAM remark LINK

• French MPs vote for a €0.055 tax on every gigabyte sent to users LINK

• Shopify Reduces Stripe Dependency LINK

• Anthropic reports $17 billion in Q3 revenue, purchase commitments of $895.4 billion in revised S-1 LINK

• Marvell to buy Nvidia-backed Lumentum in a $115bn all-stock deal LINK

• DeepSeek aims to rival Anthropic with new coding agent LINK

Trending tools

Recall for Mac: transcribes calls in real time without bots, plus a smart clipboard manager, temp shelf, music lyrics, weather, and floating desktop widgets. LINK

Orbit: blocks distracting apps on a schedule using Apple's Screen Time, adding a hold-to-open delay that makes habitual phone-checking harder. LINK

AutoShelf: automatically files invoices, sorts screenshots, and organizes photos by EXIF data like camera model, lens, GPS, and date, with customizable sorting rules. LINK

bytemail: a native macOS email client for Gmail, iCloud, Outlook, Yahoo and IMAP that keeps credentials, search and mail rules on-device while stripping trackers. LINK

Open For sales: automates support ticket handling with human handoff and enterprise security, backed by a 120-day full refund guarantee up to $2 million. LINK

Off Grid AI: a private AI assistant that runs on your own devices, keeping context, memory, and approved actions across Android, iOS, macOS, Windows, and Linux. LINK

Trending papers & reports

Security scanner showdown finds that traditional code scanners and AI security agents each catch different flaws, with Snyk hitting 82% and Claude's AI 79%, but combined they reached 101 of 108 vulnerabilities, far more than either alone. LINK

Doctor-visit prep chatbot helped patients arrive better prepared, with clinicians rating sessions ~17 points more comprehensive and patients reporting ~17 points higher readiness than a basic chatbot, while staying medically safe. LINK

AI-related layoff announcements in August and September were the lowest of the year, a16z notes, even as job postings for engineers, software engineers, and manufacturing have climbed roughly 30% since last October. LINK

See you tomorrow for a new dose of ☕️ Techpresso!