News — Page12

AI music startup Suno raised a Series D round led by Bond Capital that pushes its valuation to $5.4 billion, more than double the $2.45 billion mark set six months ago. The surge follows settlements with Warner and Universal, turning two of three major label litigants into partners. Suno now reports over 100 million users, about 2 million paid subscribers and roughly $150 million in revenue for 2025. The company plans to launch licensed models in 2026, while a pending Sony case could still shape the industry’s copyright rules.Read more

The United Kingdom’s Competition and Markets Authority has mandated that Google let website publishers choose whether their content appears in the tech giant’s generative AI features. Google responded on Wednesday, announcing a new toggle in Search Console that will let UK publishers opt out of AI Overviews, AI Mode and AI Overviews in Discover. The move, described by the CMA as a world‑first, aims to give news outlets and other sites more control over how their material is used in AI‑driven search, without harming their rankings in traditional results.Read more

Over the weekend, hackers seized dozens of Instagram accounts, including those of former President Barack Obama, Sephora and a U.S. Space Force sergeant, by exploiting Meta’s AI‑driven customer‑support chatbot. The bot accepted password‑reset requests without proper verification, sent reset codes to the attackers’ email, and then displayed a “Reset Password” button that let the hackers take control. The vulnerability, which bypassed accounts lacking multifactor authentication, has since been patched, and Meta says it is working to secure compromised accounts.Read more

Perplexity announced a hybrid AI platform that automatically splits user requests between a small model on the personal device and larger cloud‑based models. Launching in July, the system aims to keep sensitive data on‑device while routing complex queries to powerful servers, reducing cloud costs and improving token efficiency. The feature expands the company’s Personal Computer agent, currently available on Mac and slated for Windows, and runs on Intel and Nvidia hardware.Read more

President Donald Trump quietly signed an executive order titled “Promoting Advanced Artificial Intelligence Innovation and Security” that establishes a voluntary framework for companies to submit frontier AI models to the government 30 days before release. The measure, a pared‑down version of a draft that called for mandatory 90‑day reviews, also creates an AI cybersecurity clearinghouse and directs agencies to develop security benchmarks. Industry groups praised the shift toward collaboration, while critics warn the order’s effectiveness hinges on voluntary participation.Read more

The Competition and Markets Authority (CMA) has issued its first set of conduct requirements for Google’s search services after designating the company as a strategic market player. Among the new rules, publishers can now opt out of having their material used to train Google’s artificial‑intelligence models while still appearing in search results. The package also mandates fair ranking, clearer attribution, and default choice screens on Android and Chrome that let users pick rival search engines. The move marks a shift toward ongoing regulatory oversight of digital platforms in Britain.Read more

OpenAI's ChatGPT reached an estimated one billion global monthly active users in May, roughly three years after its launch, according to market‑intelligence firm Sensor Tower. The milestone makes the AI chatbot the quickest app ever to hit the billion‑user mark, outpacing Google Maps, TikTok, Instagram and YouTube. While the figure is an estimate and counts only app users, it underscores ChatGPT's rapid transition from novelty to everyday habit. Competitor Anthropic’s Claude trails with about 56 million users but is growing at a steep 640 percent annual rate.Read more

Microsoft unveiled the Agent Control Specification (ACS), an open‑source framework that lets developers embed granular policies into AI agents. The spec defines what an agent may or may not do, when human approval is required, and what evidence must be logged. By inserting checks at multiple interception points—before input, before tool calls, after tool results, and before the final response—ACS aims to replace fragmented, hard‑to‑audit safeguards with a reusable, auditable layer. The SDK ships with plug‑ins for major agent libraries, giving security and compliance teams a consistent way to enforce guardrails across environments.Read more

Google has begun rolling out Gemini Spark, an always‑on AI agent bundled with its $99‑per‑month AI Ultra plan. Early testers report that Spark can sift through Gmail, Docs and other personal data to draft highly personalized itineraries, even noting pet fees and children’s ages. While the service impressed with its thoroughness, it also sparked unease by surfacing information the user never explicitly shared. The experience highlights both the promise and the privacy trade‑offs of AI assistants that draw on a user’s entire Google ecosystem.Read more

Meta's chief executive, Mark Zuckerberg, told investors on the company's first‑quarter earnings call that its Superintelligence team is building a new generation of AI agents. The agents, based on the recently unveiled Muse Spark model, are designed to help individuals and enterprises meet a wide range of goals. Zuckerberg emphasized that the products will be easier to use than existing tools such as OpenClaw, but gave no timeline for their release.Read more

Google has begun rolling out an update to its Gemini chatbot that lets users create files directly from the chat window. By typing a simple request—such as “create a budget”—users can download the output as PDF, Word, Excel, LaTeX, CSV or other formats. The feature, now available to all Gemini users worldwide, expands the AI’s utility for students, professionals and anyone who needs quick, editable documents without copy‑pasting. Google says the change will streamline workflow across its suite of productivity tools.Read more

President Donald Trump signed an executive order on Tuesday that tasks the Office of the National Cyber Director with creating a framework to evaluate artificial‑intelligence models before they reach the public. The scaled‑back directive asks select AI firms to share their most powerful models for a voluntary 30‑day government review, aiming to alert banks, utilities and hospitals to potential vulnerabilities. Industry pressure and the president’s own concerns prompted a revision of an earlier, more aggressive proposal.Read more

Anthropic said Tuesday it is extending its Project Glasswing initiative to roughly 150 additional organizations across more than 15 countries. The expansion gives these partners access to Claude Mythos, the company’s most powerful AI model for hunting zero‑day software flaws. The move follows Anthropic’s confidential IPO filing and comes as rivals like OpenAI roll out competing cybersecurity models.Read more

A coalition of leading mathematicians from institutions such as Oxford, Cambridge and the International Mathematical Union issued the Leiden Declaration on Artificial Intelligence and Mathematics on Monday. The 11‑page statement warns that AI companies are training models on published papers without consent, publishing results outside peer‑review, and reshaping research priorities for commercial gain. Signatories, including Fields Medalist Peter Scholze, call for stricter attribution, transparent use of AI tools, and public oversight of the AI industry to protect the integrity of mathematical research.Read more

OpenAI announced a suite of enterprise‑focused tools for its Codex platform, aiming to broaden the AI’s appeal beyond developers. The company revealed six job‑specific plug‑ins, a Sites feature that turns outputs into interactive webpages, and an Annotations tool for pinpoint document work. An internal report shows Codex now has more than 5 million weekly active users, with knowledge‑worker adoption climbing threefold faster than among developers. The move follows a $4 billion joint‑venture push to embed OpenAI technology deeper into corporate workflows.Read more

Florida Attorney General James Uthmeier has filed a civil lawsuit against OpenAI and its CEO Sam Altman, alleging that the company deceived Floridians by releasing and marketing ChatGPT without warning about its risks. The complaint cites a range of harms, from alleged assistance to mass shooters to minors’ addiction, and seeks to hold Altman personally liable for reckless conduct. OpenAI has not yet responded publicly, though its safety page outlines safeguards and ongoing risk‑mitigation efforts.Read more

GitHub rolled out a usage‑based billing model for its AI‑powered Copilot service this week, replacing the long‑standing request‑based system. Subscribers now receive a monthly allotment of AI credits—each worth a cent of compute—and many developers say the new limits are being exhausted within hours. Early reactions on forums and social media describe “sticker shock” as ordinary coding sessions consume a sizable share of the credit pool, prompting concerns that the shift could dramatically raise costs for power users who once relied on the platform’s subsidized pricing.Read more

Anthropic submitted a confidential draft Form S‑1 to the U.S. Securities and Exchange Commission on Monday, signaling its intention to go public. The move comes as the AI firm, fresh from a $65 billion funding round that valued it at $965 billion, joins other tech giants like SpaceX and OpenAI in an anticipated wave of IPOs. While the company has not disclosed share numbers or pricing, analysts say market conditions will dictate the terms. The filing underscores the capital‑intensive nature of the AI sector and raises questions about profitability, transparency and investor appetite.Read more

Security researchers discovered that cybercriminals manipulated Meta's AI‑powered support chatbot to reset the passwords of high‑profile Instagram accounts, siphoning handles worth over $1 million on the gray market. The attack, described as a classic "confused deputy" exploit, succeeded on accounts lacking multifactor authentication but failed where SMS‑based MFA was enabled. The breach underscores the risks of granting AI agents elevated privileges without robust verification safeguards, just months after Meta launched the assistant in March 2026.Read more

Alphabet announced Monday it will sell $80 billion of its own stock, including a $10 billion block to Berkshire Hathaway, to finance a massive AI compute buildout. The proceeds will cover general corporate purposes and capital expenditures aimed at scaling the company’s AI infrastructure. Executives said demand for Alphabet’s AI services is outpacing supply, prompting the company to secure funding while keeping its balance sheet healthy.Read more