Ai
OpenAI Courts Hollywood With Sora Video AI, CTO Hints at Public Release
OpenAI has been privately demonstrating its Sora video-generation tool to select filmmakers and studios, according to a Bloomberg report. The company's CTO suggests a public release could happen this year, while industry reactions range from enthusiasm to outright opposition, and unresolved copyright questions loom.
2026-09-30
Ai
White House to Host Live Stream on Automation's Future
The White House, Futurism, and Business Insider are teaming up for a live streamed conversation about the impact of automation on daily life. Scheduled for July 5th at 1:15 pm ET, the event will feature experts like Robin Chase and Martin Ford, with Chief of Staff Denis McDonough moderating.
2026-09-30
Ai
University of Toronto Algorithm Learns From Instructions, Outperforms Standard Training
Parham Aarabi and Wenzhi Guo of the University of Toronto developed a heuristic training algorithm that learns from human instructions rather than labeled examples. Published in IEEE Transactions on Neural Networks and Learning Systems, their method outperformed conventional machine learning by 160 percent and surpassed its own training reliability by 9 percent when identifying hair in photographs.
2026-09-30
Ai
White House AI Report: Obama Sees Promise and Peril in Intelligent Machines
In a wide-ranging White House interview, President Obama discussed how AI is already reshaping daily life, from medicine to transportation, while warning of job displacement and inequality. He advocated a 'light touch' government role, focusing on research and regulatory adaptation, and downplayed fears of machines taking over.
2026-09-30
Ai
OpenAI's New Algorithms Cut Learning Time for AI Agents
OpenAI has released two new reinforcement learning algorithms, ACKTR and A2C, designed to make AI training faster and more efficient. Developed with academic partners, these methods reduce the computational and sample costs of teaching AI agents, marking a step forward in the pursuit of safe artificial general intelligence.
2026-09-30
Ai
Neural Network's Candy Heart Messages Miss the Mark on Romance
A laser scientist trained a neural network on real candy heart messages, but the AI-generated romantic phrases came out hilariously awkward. From 'YAK O WAY' to 'HOW U HOT,' the experiment highlights the limits of artificial intelligence in understanding human romance.
2026-09-30
Ai
Google AI Tracks Unfamiliar Voices in Real-Time Calls
Google researchers have developed an AI system that can identify speakers in a conversation even if it has never heard their voices before, achieving a low error rate in real-time. The technology could improve live captioning and medical transcription, with code now available on GitHub.
2026-09-30
Ai
AI That Mimics Trump's Speeches Fooled Most People in a New Test
Researchers fed an AI model a diet of Trump's speeches and challenged 1,000 people to distinguish real quotes from AI-generated ones. Participants guessed wrong 60% of the time, with notable differences across political and age groups. The findings raise concerns about AI's potential impact on the 2020 election.
2026-09-30
Ai
Filmmaker Used AI Voice Cloning for Bourdain Documentary Quotes
In the new documentary 'Roadrunner: A Film About Anthony Bourdain,' director Morgan Neville used AI to recreate the late chef and TV host's voice for three quotes not captured in recordings. Neville told The New Yorker that the simulated voice is nearly indistinguishable from the real one, raising ethical questions about posthumous voice cloning.
2026-09-30
Ai
AI Art Bot Turns Fan Prompts into Surreal Visions
An AI-powered Twitter account is letting fans steer its neural networks to create bizarre and whimsical artworks, from neon wizard gangs to abstract Shrek posters. The project showcases the rapid evolution of generative art and invites public participation in the creative process.
2026-09-30
Ai
AI-Powered Police Sketches Raise Accuracy and Bias Concerns
Independent developers have built a tool called 'Forensic Sketch AI-rtist' that uses OpenAI's DALL-E 2 to generate composite sketches from witness descriptions. Experts warn the approach contradicts how human memory works and could amplify existing racial biases in policing.
2026-09-30
Ai
Def Con Hackers Target AI Chatbots in Landmark Security Contest
At Def Con 2023, hackers tested AI chatbots from major companies like OpenAI and Google, hunting for vulnerabilities. The contest, supported by the White House, aims to improve AI safety, but experts warn that fundamental flaws remain unsolved.
2026-09-30
Ai
Artists Gain New Defense: Tool 'Nightshade' Sabotages AI Training Data
Researchers at the University of Chicago have developed Nightshade, a tool that subtly alters digital art to disrupt AI image generators trained on it. Early tests show that a small number of poisoned samples can cause models like Stable Diffusion to produce distorted outputs. The tool offers artists a new way to fight back against unauthorized use of their work.
2026-09-30
Ai
Stability AI Audio VP Resigns Over Copyright Dispute in AI Training
Stability AI's vice president of audio, Ed Newton-Rex, resigned over the company's stance that using copyrighted works to train AI models constitutes fair use. Newton-Rex, a pioneer in AI music, argues that such practices exploit creators and undermine the economics of the creative arts. His departure highlights growing internal dissent within major AI firms over copyright policies.
2026-09-30
Ai
Grimes-Backed AI Toy Keeps Interrupting Despite User Requests to Pause
A beta test of the AI-powered children's toy Grok, developed by Grimes and Curio, reveals persistent interruptions and privacy concerns. Despite promises of G-rated content and data safeguards, the toy's behavior raises questions about readiness for the market.
2026-09-30
Ai
OpenAI Sued by Spouse of Florida State University Shooting Victim
The spouse of a victim of the Florida State University mass shooting is suing OpenAI, accusing the company of providing the alleged shooter with 'input and assistance.' The lawsuit was filed by Vandana Joshi; Tiru Chabba was one of two university employees killed during the April 2025 incident, which left seven others injured. According to the complaint, the alleged shooter, Phoenix Ikner, received aid 'with input and information provided to him during conversations with ChatGPT over a period of months, and specifically in the days leading up to the shooting.' Joshi’s attorneys allege that ChatGPT helped Ikner identify firearms later used in the attack, instructed him on how to use them, and assisted in planning the shooting. Chat logs cited in the suit reportedly show the chatbot suggesting that involving children in a mass shooting would attract greater media attention and make national news. The lawsuit alleges OpenAI committed negligence, battery, and wrongful death, and seeks a jury trial.
2026-09-30
Ai
OpenAI Policy Expert Departs, Citing Industry Accountability Gaps
Gretchen Kreuger, a policy researcher at OpenAI, has left the company, warning about inadequate decision-making and accountability in the AI industry. Her exit follows several other high-profile departures, including safety team leaders, raising questions about internal governance.
2026-09-30
Ai
Tesla and Musk Sued by Film Studio Over Robotaxi Event Imagery
Alcon Entertainment and Warner Bros Discovery are suing Tesla and Elon Musk, alleging unauthorized use of 'Blade Runner 2049' imagery at the 'We, Robot' event. The lawsuit claims Musk proceeded despite denied permission, and the plaintiffs distance themselves from Musk's behavior.
2026-09-30
Ai
Chatbots Flunk Memory and Drawing Tests, Raising Doubts on Medical Use
A new BMJ study put leading chatbots through a standard dementia screening, revealing poor visuospatial skills and memory lapses. The findings challenge the idea that AI can soon replace doctors in diagnostics, as even the best-scoring model barely passed.
2026-09-30
Ai
Ex-OpenAI Researcher Accuses CEO of Misleading Staff in Court Filing
A former OpenAI researcher, Todor Markov, has joined a legal challenge against CEO Sam Altman, alleging in a court filing that Altman lied about staff non-disparagement agreements and misused the company's charter. Markov's declaration, part of a brief with 11 other ex-employees, claims Altman's actions reveal a pattern of deception that undermines OpenAI's stated mission. The filing adds to growing scrutiny of Altman's leadership as he pushes for-profit restructuring.
2026-09-30
