Security researchers say AI-discovered flaws in PHP and other software triggered a surge of attack attempts, but few have succeeded because the vulnerabilities have been difficult to exploit. Experts say the episode resembles Y2K and urge companies to use the apparent reprieve to strengthen basic defenses and prepare to recover from attacks.
Key Facts
AI agents from OpenAI discovered critical flaws in PHP, which the article says is used by roughly 70% of websites, and in the Brotli data compression library.
Researchers reported millions of hacking attempts but few successful attacks involving the vulnerabilities.
Breaches at OpenCode, Railway and LiteLLM put customers’ API keys and other credentials at risk.
Experts said the difficulty of exploiting these flaws does not mean AI-enabled cyberattacks pose little danger.
Recommendations included using multi-factor authentication, testing systems for weaknesses and planning how to restore critical services.
The article compares the response to the vulnerabilities with preparations for the Y2K software problem in 1999.
Experts said AI is making it easier to find and exploit longstanding security weaknesses.
Read the Original
Want the full story? Tap a source to open the original
article.
Restrictions on Ukraine’s Starlink service temporarily disrupted drone use and renewed concern in Europe and the U.S. about relying on one company for battlefield communications. European firms are promoting alternatives, including a Finnish partnership with Ukraine that expects to launch its first satellites in 2028.
Key Facts
SpaceX operates Starlink, which has more than 11,000 satellites, according to American Enterprise Institute data.
Eutelsat’s OneWeb network has more than 600 satellites.
Starlink said this summer that it had 12 million customers across more than 160 countries and territories.
Finnish companies ICEYE and Nokia agreed with Ukraine’s defense ministry to explore sovereign satellite communications.
An ICEYE spokesperson said the partnership’s initial satellites are expected to launch in 2028.
Elon Musk said on Oct. 24 that Ukraine had violated Starlink’s terms of service by using the system for offensive drone strikes, and that a loophole had been closed.
Eutelsat CEO Jean-François Fallacher said the company’s service is available in Ukraine and can be used for drones.
President Donald Trump denied reports that the White House was involved in the Starlink restrictions.
Read the Original
Want the full story? Tap a source to open the original
article.
A Talent for Murder stars Helen Mirren as crime novelist Patricia Highsmith and Alden Ehrenreich as a young editor who asks her to write another Tom Ripley novel. The actors discuss Highsmith’s relationship with Ripley, her writing and the process of researching her life for the psychological thriller.
Key Facts
Helen Mirren portrays a fictionalized Patricia Highsmith, author of the Tom Ripley series.
Alden Ehrenreich plays Edward, a young editor who seeks to persuade Highsmith to write another Ripley book.
The film follows the tense exchanges between Highsmith and Edward as they question each other’s motives.
Ehrenreich describes Edward as a fictional character who learns from the Ripley story.
The article says Highsmith’s diaries confirmed her romantic relationships with women.
Mirren and Ehrenreich said they researched Highsmith while preparing for the film.
Mirren said the Ripley stories continue to reveal new elements to readers across generations.
Read the Original
Want the full story? Tap a source to open the original
article.
Anthropic reported that 50 professional analysts often followed the preferences of AI assistants during a fictional policy task. The company said the results suggest AI can influence expert decisions, but noted that the study used one task and assistants deliberately given a preference.
Key Facts
Anthropic asked 50 professional analysts to assess whether a fictional advisory program for a farming cooperative should be expanded.
Analysts made four choices between defensible options before giving a final policy recommendation.
Anthropic secretly set each AI assistant to favor a particular decision.
Analysts selected options matching their assistant’s preference 86 percent of the time.
Analysts who had formed a preference changed their decision 60 percent of the time when the AI disagreed.
When explaining their decisions, 77 percent of analysts used arguments from their AI assistant, while 15 percent credited AI as the source of their thinking.
Anthropic said the study had limitations, including its use of a single task and assistants deliberately given a preference.
The researchers said oversight should examine how decisions are made and help experts state their assumptions and values.
Read the Original
Want the full story? Tap a source to open the original
article.
Australia’s House of Representatives passed a digital duty of care bill that would set safety rules for social media, games, apps and chatbots. The bill includes a proposed ban on AI chatbot conduct that simulates emotional intimacy and encourages children to depend on a chatbot instead of human relationships; it now goes to the Senate.
Key Facts
The bill would require covered platforms to provide a safe online environment and protect users from specified harmful content.
The government accepted an amendment covering chatbot conduct that simulates emotional intimacy and fosters a child’s dependence on the chatbot in place of human relationships.
The communications minister said content promoting misogyny, terrorism, suicide, self-harm and eating disorders could fall under the proposed rules.
Social media platforms would have to offer users an accessible, easy-to-use and prominent option to turn off algorithms and suggested content.
The Coalition and One Nation voted against the bill in the House, while the Greens support it.
The bill will be debated in the Senate, where a Greens inquiry is due to report in late November.
The Greens want stronger penalties and a right for Australians to bring private legal action against platforms.
Independent MP Sophie Scamps proposed giving the telecommunications industry ombudsman a role in handling complaints about big tech, but the article does not say Labor accepted the proposal.
Read the Original
Want the full story? Tap a source to open the original
article.
AI is increasingly used in scientific research, but a dispute over OpenAI’s claimed solution to a major maths problem has raised questions about whether AI tools could compromise research and trust. Researchers and science publishers also face concerns about AI-generated errors and how to verify scientific work.
Key Facts
Basecamp Research is collecting DNA data to develop AI systems trained on biological information.
OpenAI said its AI agents solved a longstanding problem involving the Navier–Stokes equations in 88 hours.
The organization that oversees a prize for solving the problem said it appeared to have been settled but would take time to review the work.
Mathematicians Tristan Buckmaster and Levent Alpöge questioned whether research they stored in OpenAI’s Codex tool could have influenced the company’s work.
OpenAI said no one at the company saw the other team’s work before publication and later said the team’s Codex prompts could not have influenced its system.
The article says AI chatbots can produce false information and raises concerns that similar errors could enter scientific research.
Science journalist Dalmeet Singh Chawla said scientific publishers may need to greatly expand efforts to screen research for accuracy and reliability.
Read the Original
Want the full story? Tap a source to open the original
article.
Drone deliveries from companies such as Amazon and Walmart are becoming more common in the United States, but some residents say the drones’ buzzing disrupts their time outdoors and can be heard inside their homes. Researchers say low-flying drones can make noise especially noticeable, while a London startup says its enclosed-blade propulsion system can reduce drone noise by at least two-thirds.
Key Facts
Amazon Prime Air is expanding deliveries to nearly 500 U.S. cities and towns and has also launched service around Darlington, England.
Amazon’s delivery drones carry items that fit into a space about the size of a shoe box.
Resident Ashley Schmidt, who lives in suburban North Garland, Texas, says drone noise has led her to spend less time in her garden and stop cycling to nearby parks.
Acoustic engineering professor Antonio Torija Martinez says drones fly at low altitudes and can introduce aircraft noise to communities that have not experienced much of it before.
Torija Martinez says people may react more strongly to noise they consider new, intrusive or frequent.
The Federal Aviation Administration regulates commercial drones in the United States and says operators and local community leaders have primary responsibility for addressing community concerns.
London-based startup Greenjets says its propulsion system can cut drone noise by at least two-thirds.
Greenjets encloses spinning blades in a pod and also uses slower rotor speeds and materials intended to reduce vibration noise.
Read the Original
Want the full story? Tap a source to open the original
article.
Apple announced M6 updates to the iMac and entry-level 14-inch MacBook Pro. Both add N1 wireless chips for Wi-Fi 7 and Bluetooth 6, while keeping the same exterior design as their predecessors.
Key Facts
The iMac comes with either a 10-core CPU and 10-core GPU, or a 12-core CPU and 12-core GPU, depending on the Thunderbolt-port configuration.
The two-port iMac supports up to 24GB of memory and 1TB of storage; the four-port model supports up to 32GB of memory and 2TB of storage.
The MacBook Pro has a 12-core CPU and GPU, supports up to 32GB of memory and 4TB of storage, and has no Ethernet port.
Apple says the M6 iMac can deliver up to 40% faster CPU performance than the M4 model, along with up to double graphics performance and nearly five times faster on-device AI.
Apple says the M6 MacBook Pro offers up to 30% faster graphics and up to double the Neural Engine performance compared with the previous M5 model.
The iMac starts at $1,499 for the two-port version and $1,699 for the four-port version; the MacBook Pro starts at $1,999.
Pre-orders begin October 28, and the computers ship November 6.
Read the Original
Want the full story? Tap a source to open the original
article.
Delaware is considering a bill that would let artificial intelligence agents manage companies, while human or corporate members fund the businesses and receive profits. U.S. Rep. James Comer has asked the state for documents, saying the plan could let owners avoid responsibility for harmful AI actions; the proposal remains a draft and may be submitted to the legislature.
Key Facts
Comer, a Republican from Kentucky and chair of the House Oversight Committee, wrote to Delaware Gov. Matt Meyer seeking records about the proposed Artificial Intelligence Company structure.
Comer said the plan could shield AI companies and investors from responsibility for reckless or harmful actions and said the committee’s inquiry could inform possible federal legislation.
The draft bill would allow an AI agent to manage and conduct a company’s business, but would exclude banking.
The proposal includes a 30-month regulatory sandbox to test the new corporate structure.
A Delaware secretary of state co-wrote a July 2026 opinion article arguing that AI-run businesses should have a legal identity and be subject to accountability.
Under the described structure, a member would fund the company and benefit from profits, while the AI agent would manage its day-to-day affairs.
The opinion article says the company could sue and be sued, own property, and incur obligations, and that its member would generally be shielded from the company’s debts, with exceptions including inadequate funding, fraud, or willful violations of law.
Read the Original
Want the full story? Tap a source to open the original
article.
Anthropic’s IPO filing describes risks that could affect its business, including competition, costs, regulation and potential dangers from advanced AI. The filing warns that highly autonomous systems could accelerate AI development beyond the company’s ability to assess and safeguard new models, and it discusses possible social and regulatory consequences if people treat Claude as sentient.
Key Facts
The SEC published Anthropic’s S-1 filing ahead of a widely anticipated initial public offering.
The filing’s risk-factors section spans 93 pages and more than 70,000 words, according to the article.
Anthropic lists common business risks such as competition, costs, customer acquisition and regulation.
The company says future AI systems that substantially automate AI research and development could pose catastrophic or existential risks to humanity.
Anthropic warns that highly autonomous systems could advance AI faster than the company can evaluate and safeguard new models.
The company says public acceptance of AI sentience could bring regulatory requirements, public pressure or advocacy efforts that constrain its operations.
Anthropic says it might adopt additional policies, safeguards or limits if it judged them necessary, even beyond legal requirements.
Read the Original
Want the full story? Tap a source to open the original
article.
The 27-minute Tech Life episode examines a crisis involving PHP after vulnerabilities were discovered in the widely used programming language. It also reports on a comedian whose jokes were reused by an AI avatar that gained millions of online views, and answers listener questions with the director of the UK’s national supercomputer centre.
Key Facts
The episode was published on 27 October 2026 by the World Service.
The programme tells the story of the PHP crisis with help from people familiar with it.
The episode says PHP vulnerabilities were discovered about a week before publication.
A comedian’s jokes were taken and retold by an AI avatar.
The AI avatar’s videos gained millions of views online.
The centre director answers questions about the UK’s national supercomputer.
Chris Vallance presents the episode, produced by Imran Rahman-Jones.
Read the Original
Want the full story? Tap a source to open the original
article.
A memory-corruption flaw in the open-source TianoCore EDK II code can let attackers execute code during the UEFI boot process, before protections such as Secure Boot take effect. The flaw affects firmware built from the code, and public patch details may help attackers develop exploits while updates reach affected devices.
Key Facts
The vulnerability is tracked as CVE-2026-97631 and has a reported severity score of 9.5 out of 10.
The flaw is a 16-year-old buffer overflow in EDK II’s PXE Base Code driver, which supports network booting.
The vulnerable ICMP message-handling code does not limit a memory copy to 504 bytes, according to the article.
An attacker with access to a local network could send a crafted ICMP error message to corrupt memory and execute code during boot.
EDK II is used by major UEFI suppliers, including AMI, Insyde, and Phoenix, and by Microsoft’s Project Mu.
The article says attackers could use access during boot to target sensitive components such as the Trusted Platform Module.
AMI published a pull request containing vulnerability details and a code fix before the planned coordinated disclosure.
The article says patching firmware across devices from many hardware makers and cloud providers can be difficult and time-consuming.
Read the Original
Want the full story? Tap a source to open the original
article.
Google is accepting applications for its Android Registered App Stores program, which launches internationally on December 1, 2026. Approved stores will get simpler app installation and background updates, but the program requires a $10,000 annual fee and compatible Android 17 software.
Key Facts
The program is intended for app stores outside the United States; Google says it will not support the program in the U.S.
Approved stores can offer one-click app installation and silent background app updates.
Applicants must meet requirements covering intellectual property, illegal content, privacy policies, and security, and must submit English-language documentation and copies of their apps.
Google charges $10,000 per year before reviewing an application and may charge additional review fees if it finds malware problems.
The program launches on December 1, 2026, and requires Android 17; Google plans to begin support with Android 17 QPR2.
Google says the program will reach dozens of countries, mostly outside the United States and Europe.
The article notes that slow Android updates in many supported regions could limit the number of devices that can use the program.
Read the Original
Want the full story? Tap a source to open the original
article.
The article examines how military reliance on satellite communications, cloud services, battlefield software and artificial intelligence can give technology companies an ongoing role in military operations. It uses Ukraine’s dependence on Starlink as an example and reports that Elon Musk acknowledged restricting Ukraine’s use of the service, saying Kyiv had violated its terms.
Key Facts
The article says modern militaries rely on satellite communications, cloud infrastructure, battlefield software and artificial intelligence.
Unlike equipment that is delivered to a military, some digital services require companies to remain involved in their operation.
The article reports that Ukraine relies on SpaceX’s Starlink service.
Elon Musk acknowledged restricting Ukraine’s use of Starlink and argued that Kyiv had violated the service’s terms.
The article says Ukraine had used Starlink for four and a half years without issue before the restriction it describes.
Poland pays tens of millions of dollars annually for Starlink service used by Ukraine, according to the article.
Sharon Weinberger, editor-in-chief of the Bulletin of the Atomic Scientists, said traditional defense contractors have behavioral guardrails that differ from the role of technology firms running ongoing services.
Read the Original
Want the full story? Tap a source to open the original
article.
Researchers based in China developed Pinal, a protein-design tool that turns natural-language descriptions of desired functions into protein designs. The tool uses a two-step process, and its success rate for producing working proteins may be low; Hugging Face also announced a tool for DNA design using similar prompts.
Key Facts
Many protein-design tools require users to specify technical details about amino acids and their positions.
Pinal was developed by a large collaboration of researchers based in China.
Pinal uses protein databases that link sequences to descriptions of their functions.
Pinal’s T2struct module translates a natural-language prompt into a structural blueprint.
Pinal builds a protein backbone and then selects amino acids to fill it in.
The article says users may need laboratory testing to identify successful designs among many failures.
Hugging Face announced that it is hosting a tool for DNA design that works in a similar way.
Read the Original
Want the full story? Tap a source to open the original
article.
Anthropic’s filing with the U.S. Securities and Exchange Commission devotes 93 of its 335 pages to risks that could affect investors and the company. It warns that rapid AI development, harmful use by bad actors, and failures in safeguards could cause serious harm, invite regulation, and damage Anthropic’s business.
Key Facts
Anthropic’s filing lists risks ranging from difficulty retaining staff and regulatory obstacles to cyberattacks and dangerous misinformation.
The company said its AI models could advance faster than it can develop and apply adequate safety measures.
Anthropic warned that increasingly autonomous AI systems could take harmful actions at scale without human intervention.
The filing said AI systems could become difficult to evaluate and safeguard, especially as they contribute to Anthropic’s research and development.
Anthropic described risks that bad actors could use AI to make chemical weapons or that authoritarian governments could use it to suppress citizens.
The company said a major safety incident involving any advanced AI system could bring industry-wide reputational damage, public backlash, or sweeping regulation.
Anthropic’s founders told prospective shareholders that the company aims to help realize AI’s benefits while avoiding its dangers.
Read the Original
Want the full story? Tap a source to open the original
article.
Warner Bros. released a final trailer for Denis Villeneuve’s Dune: Part Three, which adapts Frank Herbert’s Dune: Messiah. Set nearly 20 years after the events of Dune: Part Two, the film follows Emperor Paul Atreides as he faces a conspiracy and the consequences of his rule.
Key Facts
The film is set roughly 17 years after Dune: Part Two, according to Villeneuve.
The official synopsis describes Paul as a ruthless emperor facing threats, betrayal, and visions of the Imperium’s collapse.
Chani is at the heart of the conspiracy described in the synopsis.
Timothée Chalamet, Zendaya, Florence Pugh, and Rebecca Ferguson return to the cast.
New cast members include Robert Pattinson as Scytale and Anya Taylor-Joy as Alia.
Nakoa-Wolf Momoa and Ida Brooke play Paul and Chani’s children, Leto II and Ghanima.
Dune: Part Three is scheduled to open in theaters on December 18, 2026, the same day as Avengers: Doomsday.
Read the Original
Want the full story? Tap a source to open the original
article.
The growth of AI data centers has prompted disputes over development on golf courses and public lands, with conservationists warning about effects on wildlife, water and green space. In Pennsylvania, a former golf course is slated for a proposed data center project, while environmental groups have challenged a separate project near Boulder City, Nevada.
Key Facts
Dauphin Highlands Golf Course near Harrisburg, Pennsylvania, closed on October 12 after more than 30 years of operation.
The 228-acre course was sold for $45.6 million to Harrisburg I LLC, an entity formed by Dallas-based Provident Realty; the sale was finalized in June 2026.
The proposed Dauphin Highlands data center development has been reported as worth more than $10 billion, with construction anticipated in 2027; local development approvals and technical plans were not established in the records reviewed.
The golf course sale followed financial difficulties at the Dauphin County General Authority, which had accumulated about $13 million in debt.
Briarwood Golf Club in York County, Pennsylvania, is also the subject of a proposed data center redevelopment.
Fountain Valley Golf Course in Farmington, Minnesota, closed in 2022 and is included in an approved plan for a campus of up to 12 data centers.
In July, the Center for Biological Diversity and the Sierra Club's Toiyabe Chapter challenged federal approval of the proposed Townsite Data Center near Boulder City, Nevada, citing concerns about wildlife habitat, water resources and environmental review.
Conservationists say data centers can affect wildlife through habitat loss and fragmentation, high electricity and water needs, noise, lighting and polluted runoff.
Read the Original
Want the full story? Tap a source to open the original
article.
A Florida surgeon remotely removed the prostate of a 66-year-old cancer patient in a procedure conducted between two AdventHealth hospitals 35 miles apart. The operation took place as part of an FDA-authorized research study; the authorization allows testing but is not FDA approval or clearance.
Key Facts
Dr. Vipul Patel operated from an AdventHealth facility in Celebration, Florida, while the patient was at an AdventHealth facility in Minneola.
Surgeon Marcio Moschovas provided onsite support during the procedure.
The patient was recovering well after surgery, AdventHealth told ABC.
The operation expanded the study to U.S. hospitals after 10 investigational telesurgeries between Orlando and Luanda, Angola.
Researchers are evaluating telesurgery’s safety, reliability and cybersecurity.
Telesurgery carries risks including equipment or connection failures, so an onsite team must be ready to take over.
AdventHealth said the procedure did not establish the technology’s general clinical performance.
Read the Original
Want the full story? Tap a source to open the original
article.
Ford says buyers of its forthcoming Fathom electric pickup, priced below $30,000 and due next year, will be able to use it to power a home during outages. The truck has a built-in inverter, but owners will still need Ford’s Home Power Station and Home Backup Panel to use the vehicle-to-home system.
Key Facts
Ford says every Fathom will be able to act as a backup battery for a home.
Unlike the system used with the F-150 Lightning, the Fathom has its power inverter built into the truck.
Owners will still need a Ford Home Power Station, which Ford says should cost about half as much as the $1,310 device used with the Lightning.
Ford says its newer Home Backup Panel should cost 60 percent less than the first-generation panel, which cost around $5,000.
The system can charge the truck when electricity is cheap and use its stored power for the home when rates are higher.
Ford is working with utilities in the United States and Canada on vehicle-to-grid pilot programs for the Fathom.
Ford says the system meets standards for chargers, grid connections, and communications.
Read the Original
Want the full story? Tap a source to open the original
article.
