Anthropic’s junior tech hiring fell from 27% to 9%, as its AI agents came to outwork staff 6 to 1
Anthropic’s own data: junior technical hires fell from 27% to 9% and junior staff from 21% to 14% of tech headcount as AI agents outworked staff 6 to 1.
Search
23 results for “madrox”
-
-
Nine in ten Anthropic staff expect AI to make a Nobel-level discovery by 2030. Most Americans don’t
Anthropic surveyed 10,980 US adults and 1,334 of its own staff: 91% of staff expect a Nobel-level AI discovery by 2030, against 39% of the public.https://madrobot.blog/2026/10/17/anthropic-ai-expectations-gap-survey-staff-vs-americans-nobel-2030/
-
Replit’s CEO says it has dropped Claude for OpenAI because ‘ChatGPT is better now’
Replit CEO Amjad Masad says Replit no longer works with Anthropic and uses OpenAI’s models instead. A source says the Claude deal is still in place.https://madrobot.blog/2026/10/16/replit-drops-anthropic-claude-for-openai-amjad-masad-disrupt/
-
Is it illegal to make AI deepfakes in the UK? The UK deepfake law in 2026, explained
Making a sexual deepfake of an adult without consent has been a crime in England and Wales since Feb 2026. Penalties, nudify app ban and how to report.https://madrobot.blog/2026/10/16/uk-deepfake-law-is-it-illegal-to-make-ai-deepfakes/
-
Claude cracked a 25-year NSA cryptography mystery in under half an hour, Anthropic says
Anthropic says Claude traced the NSA seeds behind NIST’s P-256 and other ECDSA curves to a 1990s random number generator seeded by the clock.https://madrobot.blog/2026/10/16/claude-nist-ecdsa-curve-seeds-nsa-mystery-solved-lrand48/
-
Meta’s new business terms let it crawl your website to train its AI, with no opt-out
Meta’s Commercial Terms, effective October 30, let it crawl websites of any business using Facebook or Instagram and use the content to train its AI. -
China wants to control exports of any AI model over 2 trillion parameters
China plans export controls on 23 AI technologies, from weights of models over 2 trillion parameters to AI agents and chip design. Open source is exempt.https://madrobot.blog/2026/10/16/china-ai-export-controls-23-technologies-large-models-agents/
-
Nvidia’s RTX Spark laptops go on sale with 128GB of memory to run AI agents on your own PC
Nvidia RTX Spark laptops from ASUS, Dell, HP, Lenovo, Microsoft and MSI are on sale worldwide from $2,599, built to run AI models and agents locally.https://madrobot.blog/2026/10/16/nvidia-rtx-spark-laptops-on-sale-local-ai-agents-prices-models/
-
ChatGPT can now read your open Chrome tabs, and click around in them for you
ChatGPT’s Chrome extension can now read your open tabs in Chat and fill forms, send messages and download files in Work. Here’s how it works and its risks.https://madrobot.blog/2026/10/16/chatgpt-chrome-extension-read-control-open-tabs-work-mode/
-
Bluesky’s Attie now vibe-codes games and polls from one sentence, ready for your followers to play
Bluesky’s AI app Attie can now build games, polls, quizzes and websites from a prompt. What Attie is, who can use it and how sharing on Bluesky works.https://madrobot.blog/2026/10/16/bluesky-attie-ai-site-builder-vibe-code-games-polls-websites/
-
Is Claude Fable 5.5 out yet? The Fable 5.5 release date, rumours and what we know
Claude Fable 5.5 isn’t out yet and has no release date. What Anthropic has confirmed, the Azure listing, Polymarket odds, rumours and when it could launch.https://madrobot.blog/2026/10/16/claude-fable-5-5-release-date-is-it-out/
-
Gemini usage limits explained: the 5-hour and weekly limits, how to check yours and when they reset
Gemini limits refresh every 5 hours up to a weekly cap. How to check your usage, when Gemini resets, what each plan gets and what to do at your limit.https://madrobot.blog/2026/10/16/gemini-usage-limits-5-hour-weekly-limit-reset-how-to-check/
-
Character.AI’s buyer will inherit millions of private chats, and 39 states want them protected
39 states and DC told Character.AI’s bankruptcy court any buyer must protect users’ private chats and keep child safety rules, setting nine conditions. -
Microsoft has trained its own 2.8 trillion parameter AI model, and says it beats every open rival it tested
Microsoft AI has pre-trained MAI-Base-2, a 2.8 trillion parameter model on 16,000 Nvidia GPUs. It claims 4x to 12x DeepSeek’s training efficiency.https://madrobot.blog/2026/10/16/microsoft-mai-base-2-2-8-trillion-parameter-ai-model-deepseek/
-
OpenAI and Meta warn US judges there may never be a reliable way to spot AI fakes
OpenAI and Meta told the US courts’ evidence rules committee AI images can’t be reliably detected, as judges weigh a Rule 901(c) for deepfake evidence. -
Anthropic has let six outside AI testers inside the company, and it can’t stop what they publish
Anthropic has embedded six METR evaluators with employee-like access to its systems. METR will publish its risk findings and Anthropic can’t veto them. -
OpenAI’s GPT-6 Astra solved 49 of 50 unpublished physics research problems, at under $3 each
In TheoPhysEval, 35 physicists tested AI agents on 50 unpublished research problems: GPT-6 Astra solved 49, Claude Opus 5.5 46, Gemini 3.8 Flash 34. -
Amazon’s new $75 Bee wristband listens to your conversations, and can now turn them into shopping suggestions
Amazon’s Bee Frontier ($74.99) and Bee Eclipse ($99.99) record and summarise your conversations, and now link to Alexa+ and Amazon shopping. -
The G20’s financial watchdog wants every bank AI agent to have its own ID and a human who can stop it
The Financial Stability Board’s final AI sound practices tell banks to give each AI agent an ID, log its every step and keep a kill switch.https://madrobot.blog/2026/10/15/fsb-g20-watchdog-ai-agents-banks-kill-switch-sound-practices/
-
Anthropic told IPO investors it made $16.8 billion in three months, but its growth is cooling
Anthropic told IPO investors its Q3 2026 revenue rose about 45% to $16.8 billion and that it was profitable on an adjusted basis, but growth is slowing.https://madrobot.blog/2026/10/15/anthropic-q3-revenue-16-8-billion-45-percent-ipo-investors/
-
YouTube Music will now tell you when a song was made by AI
YouTube Music is adding an AI label to the player for songs made mostly with generative AI, using its own detection tools and music partners.https://madrobot.blog/2026/10/15/youtube-music-ai-labels-ai-generated-songs-player/
-
Grok Voice 3.0 tops voice AI rankings, and it already answers 20,000 Starlink calls a day
Grok Voice 3.0 is SpaceXSI’s new speech-to-speech model for voice agents, out now in the API at $0.08 a minute and already handling Starlink support calls.https://madrobot.blog/2026/10/14/grok-voice-3-tops-voice-ai-rankings-starlink-20000-calls-a-day/
-
Google DeepMind trained an AI on Singapore’s scams to catch fraudsters inside your chats
Google DeepMind and GovTech Singapore fine-tuned Gemma into an on-device scam detector that flags manipulative chats without messages leaving the phone.https://madrobot.blog/2026/10/14/google-deepmind-govtech-singapore-gemma-on-device-scam-detection/