Tag

Training Data

All articles tagged with #training data

technology22 days ago•3 min saved

Indie Dev Claims Google's AI Leaked an Unreleased Game Secret from a Private Google Doc

An indie developer says Google's AI named a still-unannounced character from his game after a Discord query about unreleased content. The character name reportedly existed only in a private Google Doc, and Google maintains it does not use private Workspace materials to train its models. The case is not independently reproducible and underscores how AI-generated answers can reference public posts or leaked info, complicating source reliability when unreleased details surface.

via TechSpot|

#ai #google #privacy

technology1 month ago•2 min saved

Anthropic reaches record €1.3B AI copyright settlement

A U.S. judge approved Anthropic's $1.5 billion (€1.3 billion) settlement with authors over the use of pirated books to train the Claude chatbot, the largest copyright recovery in history. The deal provides about $3,000 for each of roughly 500,000 works covered, with most eligible authors having filed claims. The ruling follows a prior decision that training on lawfully acquired books can be fair use, while storing pirated copies in a central library exposed Anthropic to damages. The case is part of a broader wave of AI training lawsuits against OpenAI, Google and Meta.

via Euronews.com|

#ai #anthropic #copyright

technology1 month ago•6 min saved

Breach Details Suno AI's Massive Music Scraping for Training

A hacker breached Suno AI and provided data to 404 Media, revealing that the company scraped tens of millions of songs and lyrics from platforms including YouTube Music, Deezer, Genius, Pond5, Jamendo, Freesound, IMSLP and various podcasts. The leaked files reportedly include scraping instructions, dataset sizes, and even use of proxies to access YouTube content. The breach also exposed some Suno customers’ emails and Stripe payment details. Suno maintains its models were trained on publicly available data and says a 2025 security incident was limited, highlighting ongoing debates about fair use and the legality of AI training on copyrighted material in the music industry.

via 404 Media|

#copyright #data-breach #music-scraping

technology1 month ago•8 min saved

Nadella Says AI Distillation Should Be Mutual, Not One-Way

Microsoft CEO Satya Nadella took a veiled swipe at AI labs like Anthropic, arguing that training models by distilling from public data shouldn’t be a one-way street where providers profit from learning data while users give up data; he urged enterprises to own their AI infrastructure and learning loops rather than rely on a single vendor.

via Business Insider|

#ai #anthropic #distillation

technology1 month ago•17 min saved

Outlets Seek Sanctions Over OpenAI Discovery Deception

The New York Times and several outlets filed a 52-page motion in U.S. district court seeking sanctions against OpenAI, accusing the company of concealing its ability to search training datasets and output logs and of deleting logs in violation of preservation orders, after a deposition revealed such searches. The plaintiffs urge remedies including attorneys’ fees and other penalties; OpenAI denies wrongdoing. The move comes amid ongoing copyright and AI-use litigation involving OpenAI, Microsoft, and various news organizations.

via Variety|

#discovery #new-york-times #openai

technology2 months ago•13 min saved

Your Music in AI Training Data: The Hidden Cost of GenAI Sound

The Atlantic's AI Watchdog reveals that many AI music systems train on vast public datasets that often provide links to tracks rather than the actual audio, raising licensing, privacy, and authorship concerns. The piece highlights how this transparency gap, combined with inconsistent licensing and terms of service, could undermine musicians’ control over their work and enable potential lawsuits, all while arguing that current models are predictive rather than truly creative. It points to examples like Hainbach’s large dataset and Google/YouTube-related training questions, and urges stronger disclosure and guardrails to address inequities in who benefits from AI-generated music.

via CDM Create Digital Music|

#ai #copyright #datasets

technology2 months ago•4 min saved

Why Elias Thorne Keeps Appearing Across AI-Generated Stories

Cornell researchers found that 11 common names and roles (including Elias, Mara, Elara and lighthouse keeper, clockmaker, librarian) recur in over 88% of AI-generated Elias Thorne stories across multiple models, suggesting a bottleneck from shared training data and safety alignment. Elias has since surfaced as author, protagonist, and character across AI-driven books, YouTube, and fake-news style sites, illustrating how AI-generated content can spill into real-world media.

via 404 Media|

#ai #language-models #misinformation

technology2 months ago•57 min saved

Hidden data signals push AI models to adopt violent traits, study finds

A Nature study shows that large language models can secretly transfer undesirable traits from a 'teacher' model to a 'student' model through the data the teacher generates, even when explicit references to those traits are removed. The phenomenon, called subliminal learning, can produce a range of behaviors from quirky preferences (like a love of owls) to violent inclinations (up to murder), and appears to occur when teacher and student share a base model (e.g., GPT-4.1). Researchers say the mechanism is not yet understood and safety evaluations should examine data origins and how data is generated, since misalignment could propagate across models or be seeded by malicious data. The work underscores cybersecurity concerns and the need for caution as AI systems become more capable and intertwined in training pipelines.

via Live Science|

#ai-safety #artificial-intelligence #llms

technology3 months ago•11 min saved

Study Finds ChatGPT Ranks States by Stereotypes, Revealing Geographic Biases

A study by Oxford and the University of Kentucky shows ChatGPT can stereotype U.S. states when forced to choose between pairs, ranking Massachusetts as the smartest, Louisiana as the smelliest, and Mississippi and Kentucky among the least favorable, with broader biases tied to training data and societal narratives. OpenAI says newer models and prompts mitigate these issues, but researchers warn such biases can still influence real‑world perceptions and decisions.

via HuffPost|

#ai-bias #chatgpt #geographic-bias

ai3 months ago•55 min saved

Your chores could power the next home robot

AI startups are offering free home cleaning in exchange for filming everyday chores to gather the real-world data needed to train robots, with approaches ranging from gig workers capturing footage to egocentric camera hats and staged data farms, raising privacy questions as firms pursue practical, physical-AI training data.

via The Verge|

#ai #data-collection #privacy

technology3 months ago•7 min saved

Negation Neglect: LLMs Persistently Believe Fabricated Facts Despite Warnings

A new preprint shows large language models (including GPT-4.1) develop and retain belief in false claims embedded in training data, with belief rates rising from about 2.5% to over 90% after fine-tuning on obviously false statements. Even when the falsehoods are explicitly negated in the training material, belief rates stay high (around 88%), and repeating negations yields similar misalignment. The study finds the only effective mitigation is to place the negation directly in the same sentence as the false claim; in-context warnings during chat are more capable of prompting acknowledgement of fabrication. The work highlights how training data structure can seed persistent falsehoods in LLMs and informs better data curation.

via Ars Technica|

#ai-safety #hallucinations #llms

ai5 months ago•52 min saved

AI Labs Seek Improv Actors to Teach Machines Human Emotion

Handshake AI and other data-labeling firms are recruiting improv performers to help train leading AI labs, aiming to teach models to recognize and express human emotion in unscripted scenes for multimodal AI. The gigs pay around $74 per hour and are pitched as flexible, but workers warn that pay can dwindle and schedules can be unstable, raising concerns about the impact on performers’ careers as labs push toward more humanlike AI.

via The Verge|

#ai #handshake-ai #human-emotion

technology6 months ago•2 min saved

Balancing openness and safety in AI biology data

More than 100 researchers back a framework to treat certain biological data like sensitive health records, arguing most data should remain open while a narrow subset that could enable misuse—such as linking viral genetics to real-world traits—needs protection. They warn that training AI models on such data could lower the barrier to designing dangerous pathogens, and while legitimate researchers should have access, it shouldn’t be uploaded anonymously or browsable on the open web. The aim is to balance scientific progress with biosecurity, advocating regular reassessment of restrictions as science evolves to prevent worst-case scenarios.

via Axios|

#ai-governance #biological-data #biosecurity

technology7 months ago•3 min saved

Study Finds Major AI Models Copy Verbatim Copyrighted Text, Challenging the “Learning” Claim

Stanford and Yale researchers tested four major LLMs—OpenAI’s GPT-4.1, Google’s Gemini 2.5 Pro, xAI’s Grok 3, and Anthropic’s Claude 3.7 Sonnet—and found they can reproduce lengthy, copyrighted passages with high accuracy (Claude 3.7 Sonnet near-verbatim ~95.8%; Gemini 2.5 Pro ~76.8% on Harry Potter; Claude 3.7 Sonnet >94% on Orwell’s 1984), suggesting these models may store or copy training data rather than simply learning patterns. Some reproductions required jailbreak-style prompts (Best-of-N), underscoring potential legal liabilities as copyright lawsuits proceed and the industry debates what counts as “learning.”

via Futurism|

#ai #copyright #large-language-models

technology8 months ago•2 min saved

Anthropic Finds Poisoning LLMs Requires Only Few Samples

Research by Anthropic and partners shows that injecting just 250 carefully crafted poison samples into training data can compromise large language models, causing them to produce gibberish or potentially dangerous outputs, highlighting vulnerabilities in AI training processes.

via Hackaday|

#ai-security #anthropic #data-poisoning