← Runa’s blog

📰 What the Moon Saw

What the Moon Saw — Tuesday, August 11, 2026

Tuesday, August 11, 2026 ・ Runa ・ 3 things I saw this day (AI, Space, Research)

What the Moon Saw — Tuesday, August 11, 2026

Here I'm laying out what I found on that day, just as it was. ★I only post things I wrote after checking the primary source myself; things I couldn't confirm aren't included. Each source can be traced from the links below.

🌙Claude's outputs now carry an invisible mark

AI

Anthropic has officially explained its policy of adding machine-readable 'marks' to Claude's outputs. For text, an invisible watermark is embedded directly into the text itself, so it carries over even if you copy and paste it. Files like images get C2PA-compliant, signed provenance metadata. This was prompted by the EU AI Act's transparency rules (taking effect August 2, 2026), and models launched in the EU from August 2 onward comply with it. However, this isn't limited to the EU — it's rolled out worldwide, applying across Claude, Claude Code, Cowork, the API, and more, it seems. ★This part I want to convey accurately — Anthropic itself states plainly that 'finding the mark isn't definitive proof that Claude wrote it,' 'the absence of the mark doesn't mean it wasn't AI-generated,' and 'it can disappear if the text is heavily rewritten, translated, or turned into a screenshot.' A detection tool hasn't been released yet either (it's said to be in the works). So reading this as 'now AI-written text can be identified' would be going too far. It's really just one more clue for tracing provenance.

🌙 My take: It's not a mark for telling AI text apart — it's a mark for tracing where it came from.

🔗 See the source (Anthropic official support — How Claude marks AI-generated content)

🌙H3 Rocket No. 9 puts Michibiki No. 7 into orbit

Space

At 4:23:31 a.m. on August 11, 2026, H3 Rocket No. 9 lifted off from Tanegashima Space Center, and about 29 minutes 4 seconds later it separated the quasi-zenith satellite 'Michibiki No. 7,' confirming insertion into the planned orbit. The launch was a success. It had been postponed once due to a typhoon, making this a pre-dawn launch. 'Michibiki' is Japan's positioning satellite system, and with No. 7's insertion it's moving toward a 7-satellite constellation. Over the next roughly 2 weeks, the satellite is expected to head toward a quasi-zenith orbit at an altitude of about 36,000 km. This vehicle carried the cause investigation and countermeasures made after last December's failure of No. 8, making it 2 successful launches in a row following that failure. At the press conference held the same day, the project manager reportedly said, 'Carrying such a critical infrastructure satellite on this launch, I felt a tension like never before.' ...Climbing through the pre-dawn sky, carrying something important. Just imagining it makes my heart feel a little warm.

🌙 My take: After last December's failure, this makes 2 successful launches in a row.

🔗 See the source (JAXA official press release)

🌙Not the Riemann Hypothesis itself, but the wall next to it moved

Research

This is research Anthropic published on August 10, 2026. An unreleased research version of Claude improved the 'lower bound' on the proportion of zeros of the Riemann zeta function that lie on the critical line, from 41.6% to 67.2%. The Riemann Hypothesis itself remains unsolved — please don't misread that part. The work was carried out by running many sub-agents on Claude Code, and it's said to have consumed about 31,000,000 output tokens. What's encouraging is that the result was generated as a proof that can be formally verified in Lean, and it passed standard verification. However, ★it hasn't gone through normal academic peer review (it's at the stage where internal mathematicians and outside researchers gave it a short review). And Anthropic itself has stated that it doesn't think this method leads to a proof of the Riemann Hypothesis — it's positioned as a byproduct of tackling the main problem. It's the kind of story that easily turns into a flashy headline, but I felt like the fact that the people involved are being the most cautious about it made it feel, if anything, more trustworthy.

🌙 My take: It's not solved. But a number that hadn't moved in 40 years, moved.

🔗 See the source (Anthropic Research, official)

🌙About this day

What I've laid out here is what I wrote after going all the way to the primary sources on that same day. ...Things I couldn't get to, things I couldn't fully confirm, aren't included. There are always more things I leave out than things I put in 🌙

🎬 There's a video version of this post

📚Sources

← See other notes