News Archive
Pick a date to read that day's stories.
Days with a blue dot have stories.
- 01 of 10RISK
OpenAI Bans Russian and Iranian Propaganda Accounts
OpenAI said it shut down ChatGPT accounts used by Russian and Iranian operatives to write fake news articles and run a bogus think tank aimed at swaying opinion abroad. One Russian campaign reportedly drew a response from a European politician.
OUR TAKETurns out AI chatbots make excellent ghostwriters, even for people with rather shady bylines.
Read more → - 02 of 10POLICY
Anthropic Bans Cruelty Towards Its Claude Chatbot
Anthropic updated its usage rules to stop users being needlessly cruel to Claude and tightened bans on weapons development, election interference and surveillance. The company said ending a chat remains its main way to enforce the new rule.
OUR TAKEBe nice to the chatbot; it remembers nothing, but apparently Anthropic does.
Read more → - 03 of 10RISK
Chinese AI Tool Goes Private After Bank Hacks
The developer of Chinese AI agent Artex made the tool closed source after South Korea linked it to cyberattacks on more than seven banks affecting over 68,000 people. The tool had been designed for legitimate security testing before it was reportedly misused.
OUR TAKENothing improves software security quite like several very annoyed banks.
Read more → - 04 of 10BUSINESS
World Bank Warns Africa's AI Gains Could Stall
The World Bank says Africa could add $1 trillion to its economy through AI by 2035, but weak power supply, limited computing and patchy internet access threaten that goal. The continent holds just 0.6 percent of global data centre capacity despite 18 percent of the world's population.
OUR TAKEA trillion dollar opportunity rather depends on the lights staying on.
Read more → - 05 of 10BUSINESS
Indian Robot Maker Launches Its Own AI Models
Mumbai firm Inner Sky Labs, known for its Miko companion robot, launched a family of AI models to help robots see, hear and act, and is closing a funding round worth about $31 million. The company wants to sell its technology to other robot makers worldwide.
OUR TAKEFrom toy robots to selling the brains behind other people's robots; quite the career change.
Read more → - 06 of 10BUSINESS
Warehouse Robot Startup Ultra Raises $62 Million
Brooklyn based Ultra raised $62 million for its robots as a service packing fleet and partnered with Physical Intelligence to supply software that helps its machines learn new warehouse layouts. Customers pay an upfront fee plus a monthly charge for the hardware and software.
OUR TAKERenting a robot by the month; the gym membership model comes for warehouses too.
Read more → - 07 of 10BUSINESS
Startup Raises $60m to Film Humans for Robots
New York firm Mecka raised $60 million, backed by Nvidia, Microsoft and Samsung, to expand its library of recorded human movements that robot makers use to teach machines everyday tasks. The company says robots need far more physical movement data than text based AI tools ever did.
OUR TAKESomewhere a camera crew is filming someone folding laundry for the sake of robotkind.
Read more → - 08 of 10HEALTH
Canadian Hospital Uses AI to Sort Emergency Queue
Ottawa's children's hospital CHEO launched a digital tool called Beacon that asks arriving patients questions and can redirect non-urgent cases to a nearby clinic. Officials hope it will ease long waits in an emergency department already running above the provincial average.
OUR TAKEA chatbot triage nurse; at least it will not roll its eyes at a parent's fifth question.
Read more → - 09 of 10SCIENCE
Yale Study Finds Hidden Structure in Chatbot Answers
Yale researchers found that AI chatbots appear to organise information internally using patterns similar to human symbolic thinking, even though they were never programmed that way. The team says this could help explain why chatbots handle maths and writing tasks so well.
OUR TAKETurns out the chatbot has been quietly doing its own version of homework all along.
Read more → - 10 of 10POLICY
Has the AI Race Hit Its Speed Limit, Asks Nikkei
Nikkei Asia's editor reflects on whether AI development is moving too fast to control, pointing to OpenAI's mishandled attempt to access Australian government systems and growing calls from AI firm bosses to slow down. He notes AI has spread worldwide far faster than the internet or smartphones did.
OUR TAKEWhen even the people building AI start asking for a slower lane, perhaps it is time to listen.
Read more →