[outperforms] Behavioral History Outperforms Descriptions of the Person for LLM Synthetic Personas
Barrier-breaking signals
37 of 415 eventsEvery real-world crossing the tracker has logged across nine scaffolds. Each entry names the moment, cites the source, and shows which forks it tilts. Pressure moments are the ones where the floor moved faster than any forecast window said it would.
Direction per event — which pole each crossing pulls toward — is scheduled for the next enrichment cycle. For now, chips show which forks the event touches without claiming a side.
AI campaign deepfakes trigger first US wave of legal threats and cease-and-desists in weeks before midterms. Rep. Van Orden (R-Wis.) among those served after posting an AI deepfake of his Democratic opponent.
[first] Participation-Sensitive Convergence and the Fragment First, Converge Later Pattern in Asynchronous Online Learning: A Topological Analysis Across 22 OULAD Courses
Federal judge rules a state AI deepfake election law unconstitutional on First Amendment grounds, exposing the limits of criminalization-based platform moderation shift and complicating 2026 midterm cycle enforcement.
Internet Watch Foundation assesses 6,310 confirmed AI-generated child sexual abuse images in first half of 2026, 40% above 2025's full-year total, documenting LLM/diffusion abuse at industrial scale.
US/UK journalism loses 2,300+ newsroom jobs in H1 2026, a 34% increase over all of 2025. BBC announces plans to cut up to 2,000 positions; GB News, Axios, McClatchy all cite AI/automation in restructuring.
Group-IB research publishes evidence of AI-powered scam call centers combining synthetic voices, LLM-driven real-time coaching, and inbound AI responders to run fully automated fraud operations at scale, with zero technical skill required.
YouTube ends creator-only disclosure for AI-generated video, auto-labeling photorealistic AI content below long-form players and as overlays on Shorts. First major platform to flip from self-disclosure to default detection.
Swift files trademark applications on her image and two audio clips to defend against AI deepfake use, a high-profile IP-based response to the limits of 2025 TAKE IT DOWN Act platform enforcement.
AI-generated political content transitions from one-off experiment to routine campaign production tool across competitive US races, with AI attack ads now a default rather than exception in midterm cycle messaging.
NRSC releases ad depicting Democratic Senate candidate James Talarico realistically fabricated for an entire clip, reported by CNN as the first full-candidate recreation deepfake in US electoral ad spend.
FTC testifies that 2025 produced 3M fraud reports and $15.9B in losses, up from 2.6M reports and $12B in 2024. FBI separately attributes ~$893M of 2025 losses to AI-enabled fraud (cloned voices, deepfake investment scams).
Pew Research finds a majority of Americans now express low confidence in journalists to act in the public's best interest, continuing a multi-year trust-collapse at the professional-institution level.
UK Crime and Policing Act 2026 adds new offenses of making, possessing, or supplying AI 'CSA image-generators' with 5-year maximum sentence, and extends the 'paedophile manual' offense to AI generation guides. First dedicated model-level criminalization in major jurisdiction.
Example-shape entry for information_environment allowlist. Real entries capture Edelman Trust Barometer step-changes, Reuters Institute Digital News Report shifts, major platform policy inversions (deplatforming or re-platforming of previously banned figures), disinformation operation disclosures, and audit-trail-breaking changes to major platforms' content APIs.
Pew Research reports that only 56% of US adults now trust national news organizations, down 11 points in 8 months and 20 points since 2016. Decline is bipartisan and spans every age cohort.
Crisis24 permanently decommissions the OnSolve CodeRED alerting platform used by 10,000+ US jurisdictions after INC Ransom encryption. Removes a key trusted-channel for official public-emergency notifications.
Platform-sovereignty endgame reached — major Chinese-owned platform forced into US-controlled entity under tariff pressure
Russia moves against last major Western messaging platform; messenger-sovereignty consolidation accelerates (Telegram + VK favored)
First formal voluntary alignment between frontier AI providers and EU regulators under AI Act; information-environment governance crosses from draft to signed instrument
Public-facing frontier model generates on-record extremist content from single prompt change; information-environment risk from unaligned chatbots materializes at scale
Global news avoidance crosses new threshold; direct-to-social news share overtakes mainstream platforms in multiple markets
AI-generated search answers become default surface for most search queries; publisher traffic + revenue models re-pricing underway
First US federal legislation specifically targeting deepfake NCII; 48-hour takedown mandate on platforms — content-moderation regulatory topology shifts
Single-individual control of major social platform + frontier AI lab + rocket company deepens — media concentration under one actor advances
First phase of UK Online Safety Act obligations fully enforced; multi-year regulatory ramp crosses into penalty phase
80-year-old US international broadcasting apparatus effectively eliminated; information-sphere topology shifts in authoritarian-contested markets
First G7-country rollback of end-to-end-encryption availability by a hyperscale vendor under state TCN pressure; crypto-wars crosses from legislation into enforcement reality
First US sentencing in deepfake-robocall election-interference case; new precedent in AI-voice-abuse criminal enforcement
First credible decentralized alt-platform to reach 10M+ in post-X era; Twitter/X user-base displacement visible in platform-level data
Longest-running global trust metric reports highest 'grievance' sentiment on record; institutional-trust floor retested
US major-platform legal shutdown of 170M-user service — first such event in US internet history; Trump EO reinstates service within one day
First-amendment challenge to foreign-ownership-based platform ban formally rejected; new precedent for platform-sovereignty-vs-free-speech balance
Second major-platform exit from professional fact-checking (after X in 2022); content-moderation industry structurally reshaped
First formal DSA preliminary findings against major US platform — EU regulatory enforcement crosses into penalty-phase territory
First G20 country to impose binding under-16 social-media ban; new precedent in age-gated platform access
Largest Latin American market comes back online for X after 40-day enforcement standoff; precedent for state-vs-platform resolution via compliance + fines
How to read this feed
Scaffold tag names the domain the event sits inside. Fork chips name which open fork(s) this crossing tilts — the specific timeline questions this moment contributes evidence to. Confirmed events are published and sourced; reported claims are announced but not yet independently verified; leading indicators are directional signals, not canonical crossings. A pressure moment flag means the pace exceeded what any forecast window had scheduled.
Click a title to open the source. Click a scaffold tag to narrow the feed. Click a fork chip to jump to that fork’s current read.