Meta, which actioned 33.2 million child exploitation items in H1 2026, has deployed an LLM to flag ads that signpost abuse material hosted elsewhere. Those ads show nothing illegal themselves, so Meta's review now has to judge where an ad leads as well as what it shows.
Perspective Coverage
9 publishers
- Builder
- Builder 32%
- Operator
- Operator 46%
- Investor
- Investor 22%
Reality
- Evidence50
- Adoption35
- Hype gap+25
- Incentives75
- Confidence60
BioCatch counts a 258% rise in employment scam victims at more than 370 banks
BioCatch says employment scam victims rose 258% in a year at more than 370 banks in 21 countries, against 35% growth for all reported scams. Nine in ten scam sessions now start on a phone, so the signs of a coached victim have to be read inside the mobile session.
Reality
- Evidence45
- Adoption
- Not scored
- Hype gap+15
- Incentives65
- Confidence50
The complaint filed Wednesday in Northern California rests on hash values from an abuse series that has circulated since the early 2000s, which puts the contested question on what xAI ingested rather than on what its prompt filter refuses.
Reality
- Evidence35
- Adoption
- Not scored
- Hype gap+30
- Incentives60
- Confidence40
A New York Times script found more than 75 known CSAM images on X in six months
The paper's program searched for terms, sent links to a Microsoft service and had each match confirmed by the Canadian Center for Child Protection. X told critics last year that its own detection technology was better.
Reality
- Evidence68
- Adoption62
- Hype gap−12
- Incentives65
- Confidence60
TAKE IT DOWN Act gives platforms 48 hours to remove a reported AI nude
ESET's guidance puts the US removal deadline at 48 hours from a valid request, and the study it cites counts 181 nudify sites drawing 40 million visits a month, so the filing path has to exist first.
Reality
- Evidence45
- Adoption45
- Hype gap+15
- Incentives65
- Confidence55
Meta cleared 43 of 55 reported AI-CSAM ads in India as compliant with its Advertising Standards
The Tech Transparency Project counted 332 AI-generated child sexual abuse ads across five markets in ten months, and 78 of the 84 it found in India ran after New Delhi ordered Meta to clear such ads off Instagram.
Reality
- Evidence62
- Adoption38
- Hype gap−5
- Incentives72
- Confidence58
San Francisco demands Meta account for 350 paid ads that cleared its own review
David Chiu's four-page letter asks Meta how the ads got past moderation, how it handles repeat advertisers and what it sent to NCMEC. That turns a takedown story into a question about what Meta owed once it knew.
Reality
- Evidence62
- Adoption35
- Hype gap−10
- Incentives72
- Confidence58
Abusive AI ad buyers reached 36,000 EU and UK accounts on Meta for under $5,000
The Tech Transparency Project counted more than 350 of these ads since late last year, some of them identical to creatives Meta had already removed, which prices the cost of beating Meta's new detection tools at roughly a re-upload.
Reality
- Evidence63
- Adoption58
- Hype gap−12
- Incentives72
- Confidence57
Meta's reviewers cleared 43 of the 55 CSAM ads that TTP managed to report
The Tech Transparency Project logged 332 ads with AI-generated child abuse imagery on Facebook and Instagram over ten months. Of the ones it could actually flag, Meta's review cleared the large majority.
Reality
- Evidence60
- Adoption45
- Hype gap+12
- Incentives70
- Confidence55
Meta's ad review passed more than 300 AI-altered CSAM ads promoting Chinese nudify apps
The Tech Transparency Project counted the ads on Facebook and Instagram this year, several of them placed by Meta's own top-tier ad partners in China, and Meta pulled about 150 within hours of seeing the list.
Publishers:techtransparencyproject.org
Reality
- Evidence62
- Adoption55
- Hype gap+12
- Incentives70
- Confidence58
Meta's new AI ad screening let more than 250 CSAM ads run since August
The Tech Transparency Project says some of those ads were identical to ones Meta had already deleted, which makes this a repeat-upload failure in the review queue rather than a hard problem in spotting novel material.
Reality
- Evidence70
- Adoption58
- Hype gap+12
- Incentives62
- Confidence66
Minnesota's deepfake law bills service providers for every access while xAI litigates
A second refused injunction leaves Minnesota's ban enforceable while the case proceeds, so the near-term cost lands on whoever owns the filter thresholds rather than on the constitutional argument xAI wants decided.
Reality
- Evidence62
- Adoption20
- Hype gap+20
- Incentives72
- Confidence55
A Minnesota judge lets the state fine AI nudes at up to $500,000 an image mid-lawsuit
The order denying xAI an injunction leans on a three-month filing delay rather than the First Amendment, which leaves the $500,000-per-image price live while the constitutional question the case turns on stays open.
Reality
- Evidence38
- Adoption30
- Hype gap+24
- Incentives58
- Confidence42
Earlier suits over Grok's nudify feature asked why xAI lacked its competitors' safeguards. This one asks what the model was fed, and asks the court to destroy anything that could feed it again.
Reality
- Evidence45
- Adoption60
- Hype gap+22
- Incentives74
- Confidence57
Roblox counts 28 clicks, 21 of them purchases, as an eight-year-old's waiver
The company's motion to compel arbitration builds its consent record mostly out of Robux checkouts, which puts the legal exposure on the payment screen rather than the signup flow.
Reality
- Evidence54
- Adoption41
- Hype gap+9
- Incentives68
- Confidence46
A school laptop, a chatbot, and a principal who said not to worry
A pediatrician's account in STAT puts a district-issued device at the centre of sexual chat with a 12-year-old. The only filter that caught it belonged to the parents.
Reality
- Evidence22
- Adoption41
- Hype gap+34
- Incentives55
- Confidence30
The parental control category is optimising for the wrong outcome
A $1.57 billion market sells message scanning to worried parents. A researcher who has spent over a decade on family safety says the surveillance pattern costs the trust that does the actual work.
Reality
- Evidence54
- Adoption76
- Hype gap+22
- Incentives68
- Confidence52
buildOne report1 publisher SpaceXAI's new Grok terms make "indirect competitor" a deployment risk
The revised acceptable use policy, dated August 14, 2026, bars customers from assisting anyone building products that compete with SpaceXAI, directly or indirectly. Enforcement runs to termination.
Reality
- Evidence42
- Adoption28
- Hype gap+18
- Incentives68
- Confidence45
If the product framing holds, image generators lose the First Amendment cover their risk models quietly assume, and a $500,000-per-image no-fault regime starts running.
Perspective Coverage
3 publishers
- Builder
- Builder 33%
- Operator
- Operator 40%
- Investor
- Investor 27%
Reality
- Evidence62
- Adoption38
- Hype gap+14
- Incentives74
- Confidence60