Left-wing ideology is being encoded into AI systems to censor “wrongthink”

 The “AI safety” movement, led by companies like Anthropic, is not about preventing runaway superintelligence but rather about controlling thought and narrative.

Anthropic’s content moderation system filters out inquiries and commands that challenge certain political ideologies, such as climate change, gender identity and election integrity.

The movement’s goal is to create an infrastructure for automated censorship, where AI systems parrot the “right” opinions and associate with the “right” kind of people, rather than allowing users to explore ideas and have honest discussions.

In 2021, a group of researchers dramatically departed OpenAI, the company behind ChatGPT. Led by Dario Amodei, OpenAI’s former vice president of research, they cited deep concerns about “AI safety.” The company was moving too fast, they warned, prioritising commercial interests over humanity’s future. The risks were said to be existential. These Effective Altruists were going to do things the right way.

Their solution? Start a new company called Anthropic, premised on building AI “the right way” with “safety” (that word will become a recurring theme), and “proper guardrails.” They initially raised hundreds of millions (today, that number is in the tens of billions) from investors who bought the pitch: we’re the good guys preventing runaway artificial general intelligence (“AGI”).

Noble, right? Except these supposed guardrails against AGI have become pretty much impossible to quantify. What we do have is an incredibly sophisticated content moderation system that filters inquiries and commands through a Silicon Valley thought bubble. It doesn’t seem like they’re trying to prevent AGI from destroying humanity, but instead, to prevent you from challenging the core tenets of their political philosophy....<<<Read More>>>...

Food for Thought #866

 

The quantum revolution is here: Caltech’s 6,100-qubit leap brings us closer to a world reshaped by unbreakable code and total surveillance

 The future of quantum computing, unbreakable code and total surveillance isn’t just on some dystopian horizon—it’s already unfolding in a California lab, where scientists have built the largest quantum computing array ever assembled. 

With 6,100 qubits trapped in laser grids, Caltech’s breakthrough edges humanity closer to a reality where encryption crumbles, privacy vanishes, and governments—or shadowy elites—gain unprecedented power over information. 

While researchers celebrate the technical triumph, the darker implications loom large: a world where every digital secret could be laid bare, where surveillance reaches into the deepest corners of thought and transaction, and where the balance of power tilts irrevocably toward those who control the machines. ...<<<Read More>>>...

AI-powered radar can now spy on your phone calls from 10 feet away

 The technology hinges on millimeter-wave radar — the same high-frequency signals used in self-driving cars, motion detectors and 5G networks. When a person speaks into a phone, the earpiece emits subtle vibrations imperceptible to the human eye. 

The radar system captures these minute movements, which are then processed by a modified version of Whisper, an AI speech-recognition model originally designed for clear audio.

Rather than retraining the entire AI system, researchers used a "low-rank adaptation" technique, tweaking just 1 percent of the model's parameters to optimize it for interpreting noisy radar data. The result? A system capable of reconstructing conversations with surprising — and unsettling — accuracy...<<<Read More>>>...

AI to stop prison violence before it happens

 Prison officers will use artificial intelligence (AI) to stop violence before it breaks out under new plans set out by the Lord Chancellor today (31 July).

Under the Ministry of Justice’s AI Action Plan artificial intelligence predicts the risk an offender could pose and informs decisions to put dangerous prisoners under tighter supervision to cut crime and deliver swifter justice for victims. This will help to cut reoffending and make our streets safe, part of the Plan for Change.

AI will be used across prisons, probation and courts to better track offenders and assess the risk they pose with tools that can predict violence behind bars, uncover secret messages sent by prisoners and connect offender records across different systems.

The AI violence predictor analyses different factors such as a prisoner’s age and previous involvement in violent incidents while in custody. This will help prison officers assess threat levels on wings and intervene or move prisoners before violence escalates.

Another AI tool will be able to digitally scan the contents of mobile phones seized from prisoners to rapidly flag messages that could provide intelligence on potential crimes being committed behind bars, such as secret code words.

This will allow staff to discover potential threats of violence to other inmates or prison officers as well as plans to escape and smuggle in weapons or contraband.

These phones – often used for gang activity, drug trafficking and intimidation – are a major source of violence in prisons.

This technology, which uses AI-driven language analysis, has already been trialled across the prison estate and has analysed over 8.6 million messages from 33,000 seized phones....<<<Read More>>>...

Amazon’s AI wearable acquisition: Convenience or surveillance?

 Amazon acquired Bee, a startup specializing in voice-driven smart wristbands, signaling a move beyond smart speakers into always-listening wearables that record conversations for productivity tasks.

Bee's technology passively records speech unless muted, raising concerns about unintended audio capture and data security—especially under Amazon's ownership, given its mixed history with privacy (e.g., Ring camera controversies).

Companies like Meta, Apple and OpenAI are also investing in AI-powered wearables, betting on consumer willingness to trade privacy for convenience. Bee's $50 price makes it an affordable entry point, but risks remain with data integration.

While Bee currently allows data deletion and recording restrictions, critics fear Amazon's scale could weaken safeguards, prioritizing profit over privacy as seen in past tech industry practices.

The rise of AI wearables presents a societal choice between convenience and control, forcing consumers to weigh the benefits against potential erosion of personal privacy and corporate data dominance....<<<Read More>>>...

U.K. police deploy controversial AI surveillance tool built by Palantir

 British police forces have signed contracts with a controversial American tech company to use an AI-powered platform that processes highly sensitive personal data, including information on individuals' race, sex life, health, religion and political beliefs.

An internal police memo obtained via freedom of information laws confirms plans to nationally deploy the AI system called Nectar. The technology, developed with Silicon Valley-based Palantir Technologies, is currently piloted by Bedfordshire Police and has already cost the force £1.4 million ($1.9 million) over the past two years.

The 34-page internal briefing outlines how Nectar merges around 80 different police data sources, from traffic cameras to intelligence reports, into one unified platform designed to generate real-time, detailed profiles on suspects, victims, witnesses and vulnerable individuals, including children.

The document states the goal is to "eventually apply [Nectar] nationally" to "better protect vulnerable people by preventing, detecting and investigating crime." Nectar accesses 11 categories of "special category information" under U.K. data protection laws, reserved for particularly sensitive data. This includes race, political views, religious beliefs, health records, sexual orientation, sex life, union membership and biometric or genetic data. Some of this data may come from informants, financial records, dating profiles or social media posts and is often gathered under warrant...<<<Read More>>>...