✨ MYSTERY AI LINK
This link leads to one of the most interesting things I’ve seen in AI recently.
THE BIG STORY
What’s cookin’? Anthropic is expanding the invisible watermark in Claude’s writing to older models, meaning text generated with Claude can more easily be identified. Its support guide explains that supported models leave a pattern in their word choices, rather than add hidden characters or markings. Anthropic says this doesn’t change the meaning or readability of a response and contains no information identifying the user. It’s worth noting that it can indicate that Claude likely helped write something, but it cannot prove who wrote it or whether someone followed the rules. As of now, detection is limited to eligible organizations in a private preview.
🤔 Hunter’s take: This could make checking where writing came from more useful, but we need to be careful about what a result means. Someone might use AI to write an entire essay or to revise a draft they wrote themselves. A signal that Claude was involved doesn’t settle whether the person was honest about that involvement.
DAILY POLL
Should AI-written text carry an invisible watermark?
ALSO ON THE MENU
What’s cookin’? Meituan’s LongCat team has released LongCat-2.5, an AI model designed to work with images and carry out longer software projects. The team says it can use tools and work through multiple steps to build or improve software. Its visual abilities are intended for jobs such as turning designs into working pages and helping with creative or 3D projects. It is currently available in the API changelog.
🤔 Hunter’s take: The useful test is whether an AI can stay on track when a project gets messy. Producing a good first draft is one thing, but understanding a design, fixing mistakes, and finishing the job takes more persistence. I’d judge this release by how much work people still have to redo before they can use the result.
🛠️ NEW AI TOOLS 🛠️
🎧 SUN
Turn a prompt or PDF into an audio lesson or podcast you can listen to on a walk or commute.
🎙️ HoldTalk
Hold the fn key on your Mac and speak to type into any app. A free tier includes 2,000 words a week.
PROMPT OF THE DAY:
I am torn between [option A] and [option B]. Ask questions that reveal what I value, what I fear, and what I may regret. Then compare the options using near-term benefits, long-term consequences, reversibility, and opportunity cost.
HAS AI REACHED SINGULARITY? +0%
Holding at 76.5%, with no change today. Longer software tasks are a useful test of AI progress, but a new release alone doesn’t show that those systems can reliably handle open-ended work.
The meter reflects my editorial judgment.
HOW ARE THE BIG AI PLAYERS DOING?
(Click a ticker for more info)
Have a Big Question about AI?
Here’s your chance to get a straightforward answer, from an expert in the field.
What do you think of today's newsletter?
I read all feedback. Thank you!
Hi — I’m Hunter, a PhD candidate whose work has appeared in major academic journals and popular tech outlets. I founded FryAI to make staying ahead of AI clear, accessible, and fun.