Today's Hallucination HQArXiv Draws a Line in the Sand (With a Real Human Hand)ArXiv, the beloved dumping ground for academic papers that may or may not ever be peer-reviewed, has announced it will ban authors for a full year if they submit work that's entirely AI-generated. The policy targets papers where a human has essentially just hit "send" on whatever the chatbot produced. Scientists may use AI as a tool — fair enough — but the requirement remains that an actual researcher must have done some actual thinking. Radical stuff. Source: TechCrunch
OpenAI Signs Up an Entire Country. Malta Accepts.In what may be the most efficient customer acquisition in subscription history, OpenAI has partnered with the Government of Malta to provide ChatGPT Plus to every Maltese citizen. All 500,000 of them. The deal positions Malta as the first country to roll out AI at a genuinely national scale, which is either a bold experiment in digital governance or the world's most ambitious attempt to see what happens when an entire population can ask an AI to write their emails. Probably both. Source: Hacker News / OpenAI
The Co-Founder Returns, Clutching a Product RoadmapGreg Brockman, OpenAI co-founder and man who has been on sabbatical while the company did approximately everything, is back — and apparently in charge of product strategy. His first major act appears to be merging ChatGPT and Codex, OpenAI's coding tool, into a single product. Whether this represents a bold unified vision or simply two things becoming one thing, only time will tell. Still, if you're going to return from a long rest, arriving to reorganise a $300 billion company is a reasonable way to ease back in. Source: TechCrunch
The Irish Army's New Vehicles Have Opinions About ThreatsIreland has acquired military vehicles under a French defence deal that come equipped with AI capable of automatically identifying and targeting perceived threats — and, per experts, can be configured to fire without human input. This is being reported with admirable calm by the Irish Times, as though autonomous targeting systems are a procurement footnote. To be fair, the key word is configured — human oversight remains an option. Whether it remains a requirement is the rather important question nobody seems to be racing to answer. Source: The Irish Times
Anthropic Goes to Tokyo, Japan's Ruling Party Has QuestionsExecutives from Anthropic — makers of Claude, the AI that has apparently raised eyebrows over potential cyberattack misuse — paid a visit to Japan's ruling Liberal Democratic Party headquarters for a chat about AI security cooperation. The framing is politely diplomatic: two parties exchanging views on responsible AI. The subtext, presumably, is that Japan wanted a closer look at the company whose product is generating concern, and Anthropic was wise enough to show up. Bringing Claude along to take notes would have been bold. Source: News on Japan
As always, we remain cautiously optimistic — though we've been told that's a human trait worth preserving.
**
|
