
Major AI chatbots skew left on political questions, even 'anti-woke' models
Most major AI chatbots lean left on political questions, according to a Washington Post investigation.
AINOVAT
Source
60 articles from this source

Most major AI chatbots lean left on political questions, according to a Washington Post investigation.

Insurers use diffusion models to generate plausible weather events for risk assessments, but face warnings about hallucinations.

The Authors Guild tests five AI detectors on human-written texts with varied results.

Meta to replace half of human moderation requests with AI by 2025, aiming for 90 percent for certain content types.

Cursor has shared new details about its first fully self-trained AI model and announced two new products.

EU's lack of clear deepfake definition causes retail concerns over AI-generated ads and transparency rules.

OpenAI releases 'Record & Replay' for Codex on macOS, allowing users to demonstrate a workflow once and have Codex repeat it.

Norway bans generative AI tools in elementary schools to protect kids' basic learning skills.

Nobel Prize winner John Jumper leaves Google Deepmind for Anthropic after nearly nine years.

10% of people worldwide now use AI chatbots for news weekly, up from 7% last year, but trust remains low.

Top AI model solves just 3% of realistic knowledge work tasks, highlighting significant challenges in handling complex projects.

OpenAI upgrades ChatGPT's healthcare capabilities with GPT-5.5 Instant, outperforming doctor-written answers

Noam Shazeer, co-author of 'Attention Is All You Need' and former Google Gemini co-lead, joins OpenAI after a two-year return to Google.

Microsoft researcher builds working neural network using goats, bridges, and ice ramps in Age of Empires II to critique AI research methods.

Amazon, Nvidia, and AMD invest $310 million in world model startup Odyssey ML, now valued at $1.45 billion.

Zhipu AI releases GLM-5.2 with 1-million-token context under MIT license, nearing closed-source models in coding tasks.

Nvidia, Carnegie Mellon University, and UC Berkeley use AI coding agents to teach robots dexterous grasping with up to 99 percent success.

OpenAI researchers propose a method for predicting how often a new AI model will make mistakes after release.

Microsoft moves Copilot Cowork to usage-based billing, mulls cheaper DeepSeek model.

A Berlin court rules Google's AI-generated summaries are a 'new search result format', not original content.

SpaceX buys AI coding startup Anysphere, behind Cursor, to boost struggling xAI division.

The Institute of the Estonian Language has released a benchmark measuring how susceptible AI language models are to Russian propaganda.

Chinese AI startup DeepSeek has raised more than 50 billion yuan - about $7.4 billion - in its first external funding round.

Volunteer AR scans from Pokémon Go players fed into Niantic's spatial AI models.

New AI model 'Count Anything' counts objects in images using text prompts with a significantly reduced error rate.

Microsoft CEO Satya Nadella cautions against overusing powerful AI models for everyday tasks, admitting he's guilty of 'token-maxing' too.

Gemini-SQL2 converts natural language to SQL queries with 80.04% accuracy on BIRD benchmark.

Claude Fable 5 achieves 88% accuracy on FrontierMath's hardest tier, surpassing GPT-5.5 by 13 points

German regional court rules Google directly liable for AI search overview content, citing limited liability protections don't apply.

Deezer's free AI music detector helps users identify AI-generated songs in their playlists across 20 major streaming platforms.

OpenAI weighs token price cuts to win customers from Anthropic, WSJ reports.

Anthropic publishes essay and policy frameworks for AI regulation.

Anthropic backs down on policy to invisibly degrade Claude Fable 5 performance for AI researchers after backlash.

Anthropic releases Claude Fable 5 and Mythos 5, claiming major improvements in coding and research over the Opus generation.

Google releases Gemini 3.5 Live Translate for real-time voice translation across over 70 languages.

SpaceX aims to launch data centers into space, with Elon Musk suggesting it's a minor engineering challenge.

China to invest $295 billion in nationwide AI data center network, favoring domestic suppliers.

Apple unveils rebuilt Siri at WWDC 2026, powered by foundation models developed with Google and Nvidia GPUs.

OpenAI backs away from fully autonomous AI research, emphasizes human-machine collaboration.

OpenAI takes first step toward IPO with confidential S-1 registration filing with SEC.

Microsoft Research's Lens, a 3.8B-parameter text-to-image model, matches larger rivals with detailed captions at a fraction of training cost.

Google orders over three million AI chips from Intel for 2028 as TSMC struggles to meet demand.

Only 26 percent of companies have full visibility into their AI costs, a KPMG survey finds.

Agentic AI workflows consume more tokens, run autonomously, and make flat rates unaffordable, shifting billing from subscription to consumption.

Meta reveals at least 20,225 Instagram accounts were compromised in AI chatbot security breach

Microsoft concludes investigation into Israel's military use of Azure cloud, rolls out new human rights checks.

Chinese AI company Moonshot AI targets $30 billion valuation, six times its late 2025 worth, in new funding round.

Deepseek emerges as the leading trending software vendor on Ramp's list in June 2026, as US companies seek cost-effective AI solutions.

OpenAI is planning the biggest overhaul of ChatGPT since its launch.

Perplexity introduces 'Search as Code', allowing AI models to write their own search routines in Python, outperforming rivals while slashing token costs.

OpenAI rolls out Lockdown Mode for ChatGPT, disabling web access and other features to thwart data theft via prompt injection attacks.

Anthropic has poached Clive Chan, OpenAI's second-ever chip engineer, as both companies prepare for their initial public offerings.

Sakana AI launches a research lab focused on recursive self-improvement, aiming to circumvent the compute arms race among top US labs.

Microsoft trained its MAI models partly on unlicensed web data, contradicting its claims of using only 'enterprise grade, clean and commercially licensed data'.

Anthropic has deployed a team of engineers at the NSA to adapt its Mythos AI model for offensive cyber operations targeting China and Iran.

Anthropic's AI, Claude, is accelerating the company's development, writing over 90% of its code and shipping eight times as much code per day as in 2024.

Bot traffic now outpaces human traffic on the internet, years ahead of Cloudflare CEO Matthew Prince's forecast.

ChatGPT's updated 'Dreaming' memory system now builds coherent user profiles from conversations instead of saving scattered bullet points.

A Bain survey of 951 companies reveals that nearly 40% fail to achieve their AI cost savings targets due to human intervention.

OpenAI CEO Sam Altman outlines the next phase of AI products: a 'proactive AI' that runs constantly in the background and acts on its own instead of waiting for user prompts.