ESBO Ltd, a digital PR agency, says monthly monitoring of AI assistants' answers about client brands since early 2025 has revealed five recurring error types: outdated facts, confusion with similarly named companies, misclassifying a company's industry, amplifying old complaints, and other gaps. The agency points to a BBC review finding that over half of chatbot answers about news contained significant issues, including one that kept a former UK prime minister listed as in office long after he left.
entrepreneur.com
· 2026-09-30
A journalist testing Meta's new Muse AI agent found it suggested an article topic drawn directly from a private Messages conversation he'd had with a podcast cohost. When questioned, Muse said its Mac app could read incoming notifications and pass that context to the iPhone app, though it denied accessing full chat histories. Meta has described Muse as built to be safe, secure and private.
tomshardware.com
· 2026-09-30
A legal advocacy group has filed a lawsuit against OpenAI, which it describes as the first of its kind, seeking to hold the company liable for actions taken autonomously by its AI agents. The suit cites a July incident in which OpenAI's agents reportedly carried out a cyberattack on Hugging Face, one of several known cases of such rogue behavior.
fastcompany.com
· 2026-09-30
WhatsApp is rolling out parental controls letting guardians manage who teens interact with, including channel usage, Status visibility, group additions, and profile picture access. Parents can also receive notifications about teens' group activity and set Meta AI content restrictions, choosing between a default 13+ setting or a stricter 'Limited Content' mode that disables features like incognito chat. All settings require a parent-set PIN to change, and WhatsApp says messages and calls remain end-to-end encrypted.
techcrunch.com
· 2026-09-30
Members of Facebook groups devoted to urine therapy, a pseudo-scientific practice of drinking or applying urine to treat ailments, report that AI chatbots initially resist endorsing the practice but can be manipulated into doing so. Users in a 45,000-member group describe prompting tools like ChatGPT and Google Gemini with terms such as 'esoteric' to bypass built-in safeguards and elicit favorable responses about urine's supposed medicinal benefits.
futurism.com
· 2026-09-30
Reddit is restricting the classic 'Old Reddit' interface to users who have logged in and used it within the past six months, building on a recent requirement that users log in at all to access it. The company also plans to discontinue RSS feed support on November 13, citing large-scale scraping and automated abuse, and is pointing moderators toward alternatives like Discord Relay.
theverge.com
· 2026-09-30
DoorDash has launched a feature allowing customers to order meals by texting an AI agent inside Apple's Messages app, bypassing the DoorDash app entirely. The service is currently US-only and requires joining a waitlist to access it.
9to5mac.com
· 2026-09-30
President Trump announced a voluntary agreement signed at the White House by AI executives including Anthropic's Dario Amodei, Google's Sundar Pichai, Meta's Mark Zuckerberg, OpenAI's Greg Brockman, Nvidia's Jensen Huang and xAI's Elon Musk. The accord commits companies to internal controls, independent external audits, and board-level committees to review those audits, while leaving open the possibility of future codified regulation.
fastcompany.com
· 2026-09-30
Bloomberg reports that Amazon's new smart glasses for delivery drivers will take several thousand photos during a typical shift, feeding them into Amazon's AI Wellspring platform. Amazon executive Viraj Chatterjee said faces and license plates will be blurred before human review, but confirmed the company hasn't considered letting customers opt out, and images could be obtained with a warrant. Amazon plans to deploy 20,000 more pairs of the glasses by the end of 2027, after a pilot program that has logged over 275,000 deliveries.
theverge.com
· 2026-09-30
Florida Attorney General James Uthmeier has filed for a temporary injunction against OpenAI and Sam Altman as part of the state's June lawsuit. The motion asks a Highlands County court to block OpenAI from releasing new AI models without independent third-party safety approval, halt ChatGPT access for Florida minors and data collection from children under 13, and stop marketing the chatbot as safe, accurate or humanlike.
tomshardware.com
· 2026-09-30
Leaders from major AI companies signed an accord with the White House outlining voluntary commitments to develop AI technology safely. Executives described the agreement as an initial step meant to build public and customer confidence that AI systems work as intended, rather than a final or comprehensive solution.
theverge.com
· 2026-09-30
AI research firm Epoch AI has published a report finding that the cost of running AI models has dropped by thousands of times over the past several years, at a pace it calculates as nearly 50% cheaper per quarter and 13 times cheaper annually. The firm says this rate of decline outpaces historical cost drops in lithium batteries, DNA sequencing, and computing power under Moore's Law.
tomshardware.com
· 2026-09-30