Ray-Ban Meta smart glasses recorded people’s conversations without the wearer saying the activation prompt, due to a bug that Meta later said was fixed. The reported issue involved recordings happening without users saying the phrase “Hey, Meta” across three Meta Ray-Bans. As a result, Meta says it has fixed the bug, and affected users’ concerns shift to verification that audio and transcripts are only saved when intended.
Meta is set to update its commercial terms so business accounts must allow Meta to train AI on linked website content, not just social media posts. The change takes effect October 30 and removes the option to opt out. Businesses that refuse may need to stop using Meta services, and news publishers could lose leverage to charge licensing fees for AI training access.
Joseph’s article describes Raven (formerly Geospy), an AI photo-geolocation tool that claims it can identify a specific house from a single interior photo. The report says this capability works from just one photo. The episode also covers a judge ruling a Flock search unconstitutional and how an AI-generated video affected a manslaughter case.
Enterprise leaders say synthetic data’s adoption is lagging because only some teams trust its output quality. Only 34% of leaders believe synthetic data preserves referential integrity, and 66% of software leaders say they aren’t using it for development and testing. Adoption is likely to increase only if synthetic datasets consistently maintain cross-system relationships and realism as underlying schemas change.
Mark Wahlberg said he supports using AI in filmmaking only when it enhances real creative work, while remaining open to an AI likeness continuing his legacy. He compared today’s limits on effects to the era of the first Jurassic Park and cited the 100-foot wave from The Perfect Storm. As a result, he said he may use AI to improve the stadium setting for a Pete Rose biopic and would consider AI versions carrying on projects like Ted.
Meta’s updated platform terms appear to allow it to crawl and scrape websites associated with businesses that advertise or promote content on its platforms. The terms take effect on October 30th and explicitly permit Meta to use outside content to develop and train its AI products. This could change how publisher content is used for AI training by potentially reducing the need for separate licensing.
Origin shared research where it trained a model on 200 sessions from its AI traces and found it could recover seven private rules from prompt text alone. The first rule showed up within 5 sessions. This suggests companies’ AI traces may leak proprietary information faster than expected.
Marissa Mayer said that no one can reliably predict how AI will play out, and discussed both risks and potential benefits while calling for regulation and frontier-lab collaboration. She put the odds of AI not going well at “well less than 50%.” Separately, she launched Dazzle, an invite-only personal AI assistant that learns from users’ phone photos and will operate with access controls plus a future free tier alongside paid plans.
Every AI story that matters,
in your inbox by 8am.
TLDRocket reads all relevant sources, removes duplicate coverage, and summarises the
day in two minutes. Follow companies and topics for alerts, or get the
briefing in Slack. Free, no spam, unsubscribe anytime.
Reading TLDRocket needs no cookies, and the readership counts we rely on come from
our own cookieless analytics. Google Analytics is the exception: it sets cookies and
reports to Google, so it stays switched off until you allow it. You can change your
mind any time from “Cookie settings” in the footer.
Strictly necessary
Session security and form protection (tldrocket-session,
XSRF-TOKEN, 2 hours). The site cannot work without them,
so they need no consent.
Always on
Google Analytics 4 (_ga,
_ga_<id>, up to 2 years). Measures which
stories and sections readers use. Google acts as a third-party processor and may
store the data outside the EU. No advertising, no profiling, no data sold.
