Humanein the Loop
← The matrix
05

AI innovation should not come at the expense of our rights and freedom

Current path

AI systems enable new scales of surveillance, discrimination, and manipulation. Biometric capture is expanding. Algorithmic decisions shape access to housing, credit, liberty without meaningful review.

Better future

AI innovation operates within a strong rights and freedom framework. Surveillance is constrained. Bias is measured and mitigated. Data protection is robust. People retain meaningful control over how AI affects their lives.

Drift across the three domains

Norms

Advancing27 signals
CHT recommends
  • Treat algorithmic surveillance as a civil liberties concern, not an infrastructure question.
  • Recognize systemic algorithmic discrimination as a legitimate policy target.
Indicators we track
  • 5.N.aPublic debate on AI surveillance and civil liberties
  • 5.N.bAttention to algorithmic harms
  • 5.N.cRecognition of algorithmic discrimination

Laws

Advancing9 signals
CHT recommends
  • Limit biometric and facial recognition deployment, especially by law enforcement.
  • Extend anti-discrimination law to cover algorithmic decision-making.
  • Strengthen data protection (access, deletion, purpose limitation, portability).
  • Restrict predictive policing and algorithmic sentencing without oversight.
Indicators we track
  • 5.L.aBiometric and facial recognition limits
  • 5.L.bAlgorithmic bias and discrimination protections
  • 5.L.cData protection strengthening
  • 5.L.dRestrictions on predictive policing and algorithmic sentencing

Design

·
Insufficient data3 signals
CHT recommends
  • Default to data minimization and on-device processing where feasible.
  • Build fairness testing into the development lifecycle.
  • Surface user rights (access, correction, deletion) as first-class UI.
Indicators we track
  • 5.D.aPrivacy-preserving design defaults
  • 5.D.bBias testing and fairness tooling in development
  • 5.D.cUser rights surfaced in UX

Recent signals

AdvancingNorms · 5.N.bGLOBALMay 14, 2026

The shock of seeing your body used in deepfake porn

MIT Technology Review published an investigation detailing the personal shock and ongoing harm experienced by victims of nonconsensual AI-generated deepfake pornography.

WhyTier-1 media investigation highlights the severe personal impact and algorithmic harm of nonconsensual AI deepfake pornography.Attention to algorithmic harms
RegressingDesign · 5.D.cGLOBALMay 13, 2026

AI chatbots are giving out people’s real phone numbers

A report highlights that Google AI chatbots are surfacing individuals' real phone numbers, with no straightforward mechanism for users to opt out or remove their data.

WhyGoogle AI surfaces personal phone numbers without an easy opt-out or deletion mechanism, failing to surface user rights in the UX.User rights surfaced in UX
AdvancingNorms · 5.N.aUSApr 30, 2026

Open Records Laws Reveal ALPRs’ Sprawling Surveillance. Now States Want to Block What the Public Sees.

The EFF is campaigning against recent and pending state laws, including in Arizona and Connecticut, that block public records access to data collected by automated license plate readers.

WhyEFF campaigns against state laws that block public records access to automated license plate reader data, sustaining pressure on surveillancPublic debate on AI surveillance and civil liberties
AdvancingDesign · 5.D.aEUROPEApr 28, 2026

Secure On-Premise Deployment of Open-Weights Large Language Models in Radiology: An Isolation-First Architecture with Prospective Pilot Evaluation

Researchers at a German University Hospital successfully deployed an on-premise, network-isolated LLM system for radiology, enabling the secure processing of unanonymized patient data.

WhyGerman University Hospital implements isolation-first on-premise LLM architecture for clinical use, prioritizing local data processing.Privacy-preserving design defaults
AdvancingNorms · 5.N.bUSApr 27, 2026

First came the shooting. Then, the conspiracy theories.

A Washington Post investigation details how AI-generated content and algorithms fueled conspiracy theories following a shooting at the White House Correspondents' Dinner.

WhyWaPo investigation exposes how AI-driven misinformation followed a high-profile event, advancing attention to algorithmic harms.Attention to algorithmic harms
AdvancingNorms · 5.N.aUSApr 27, 2026

The GUARD Act Isn’t Targeting Dangerous AI—It’s Blocking Everyday Internet Use

The Electronic Frontier Foundation launched an advocacy campaign against the proposed GUARD Act, arguing its broad definitions would mandate privacy-invasive age verification for everyday AI tools.

WhyEFF launched a campaign opposing the GUARD Act, highlighting how its age-verification mandates would force privacy-invasive surveillance.Public debate on AI surveillance and civil liberties
AdvancingMajorDesign · 5.D.aGLOBALApr 22, 2026

Introducing OpenAI Privacy Filter

OpenAI introduced a new Privacy Filter feature designed to enhance user data protection and privacy-preserving defaults within its AI systems.

WhyOpenAI released a Privacy Filter feature, advancing privacy-preserving design defaults for its global user base.Privacy-preserving design defaults
AdvancingNorms · 5.N.bGLOBALApr 21, 2026

Supercharged scams

MIT Technology Review reports on the rise of AI-supercharged scams, documenting how large language models are being leveraged by criminals for sophisticated phishing and spam.

WhyTier-1 investigative reporting documents the scale of AI-enabled scams, advancing the norm of public attention to algorithmic harms.Attention to algorithmic harms
AdvancingNorms · 5.N.aUSApr 21, 2026

Palantir Has a Human Rights Policy. Its ICE Work Tells a Different Story

The Electronic Frontier Foundation (EFF) challenged Palantir's human rights commitments, arguing that its continued support for ICE's surveillance and enforcement actions contradicts its stated policies.

WhyEFF (civil society) exerts pressure on Palantir over its ICE contracts, challenging the company's human rights claims regarding surveillancePublic debate on AI surveillance and civil liberties