Category: Biz Tech

Slovenia’s. si internet domain has recorded an extraordinary registration surge after US President Donald Trump moved to replace “artificial intelligence” with “super intelligence”, turning the country-code suffix into a sought-after shorthand for the technology sector. Register. si, Slovenia’s national domain registry, recorded 46,066 new. si domains in September, compared with 3,515 in August. Its published statistics showed more than 250,000. si domains registered by early October, a sharp expansion from fewer than 200,000 immediately after Trump’s United Nations announcement. The rush accelerated

Cloudflare has released two open-source artificial intelligence decision models, Clef and Clef-flash, mounting a direct challenge to TypeSafe AI’s Jev in the emerging market for fast, structured decisions inside software and autonomous-agent workflows. The company made both models available through Workers AI on October 1 and released their weights on Hugging Face under the Apache 2.0 licence. Clef has 27 billion parameters, while the latency-focused Clef-flash has 9 billion. Both accept a state and typed questions and return probabilities for permitted

X Internet Unlimited Company and X Corp have launched High Court proceedings against two Bitcoin-focused content creators and unidentified account operators, alleging they fraudulently obtained at least £207,384 through the social platform’s former Creator Revenue Sharing programme. The claim, filed on September 17 in the Business and Property Courts of England and Wales, names Vivek Kumar Sen and Zamyang Sherpa, both described in court papers as residents of Preston, Lancashire, alongside “persons unknown” alleged to have operated, used or controlled associated

Anthropic has introduced Mods for Claude Code, giving developers a way to alter the coding agent’s behaviour, interface and internal event handling through JavaScript or TypeScript functions running inside the product. The system, released on October 1 with Claude Code version 2.1.287, extends the tool’s existing plugin framework. Mods can rewrite prompts before they reach the model, block or modify tool calls, approve or reject permission requests, redact information from tool output and replace or augment parts of Claude Code’s interface. Anthropic

Google executive James Manyika has said artificial intelligence companies cannot be solely responsible for policing systems, arguing that regulation and public oversight are necessary alongside industry safeguards. Manyika, Google's senior vice-president for Research, Labs, Technology & Society, said responsibility for managing AI's risks must be shared among developers, governments and society. His comments come as debate intensifies in the United States over whether voluntary commitments by leading technology companies provide sufficient protection. Speaking in an interview released on Friday, Manyika rejected the

Artificial intelligence companies are putting friendly faces on increasingly powerful software, turning personal agents into rounded, wide-eyed characters as the industry tries to make autonomous technology feel approachable. OpenAI this week unveiled Dots, persistent AI agents with their own cloud computers and access to connected apps. Users can personalise the agents, including giving them names, while OpenAI is positioning them as assistants capable of carrying out extended tasks rather than merely answering questions. The launch follows Meta’s introduction of Muse in September.

US President Donald Trump has secured a voluntary safety accord with six major artificial intelligence companies, committing them to layered internal controls, independent audits and board oversight of their most powerful systems. The agreement, signed after a White House meeting on Tuesday, covers OpenAI, Anthropic, Google, Meta, Nvidia and Elon Musk’s xAI. Trump described the commitment as “morally binding”, while the document itself leaves compliance to the companies and creates no penalties or government enforcement mechanism. The White House Accord on Super

Attacks targeting artificial intelligence systems have become the biggest cybersecurity preparedness gap for organisations, PwC said on Thursday, as its global survey exposed weaknesses in skills, accountability and data protection. Half of security leaders placed attacks on AI systems among the threats they were least prepared to address, ahead of cloud threats at 40%, third-party breaches at 34% and ransomware at 33%. The findings were published on 1 October in PwC’s 2027 Global Digital Trust Insights survey. The study gathered responses from

Foreign investors are taking a growing role in financing North America’s artificial intelligence infrastructure build-out as technology companies tap global pools of capital to meet spending approaching $1 trillion, according to JPMorgan Asset Management. Charles Wu, head of Asia-Pacific alternatives institutional client strategy at JPMorgan Asset Management, told the SuperReturn Asia conference in Singapore on Wednesday that banks and institutional investors from Asia, the Gulf and other regions were increasingly supplying capital for computing assets concentrated in North America. “Compute training assets

OpenAI is seeking at least $30 billion in fresh capital at a valuation of about $1.4 trillion before the new investment, as the ChatGPT maker turns again to private markets after delaying plans for an initial public offering. Discussions with prospective investors remain at an early stage and terms could change, people familiar with the fundraising have said. Investor demand is driving the potential transaction as OpenAI records rapid revenue growth, while the proposed financing would provide additional capital during a

Anthropic’s confidential initial public offering prospectus has revealed a nearly $42 billion net loss for 2025 despite revenue of about $4.6 billion, alongside heavy dependence on two customers that each supplied 12% of annual sales. The disclosures offer the clearest financial picture yet of the Claude developer as it prepares for a possible stock-market listing. Revenue expanded roughly twelvefold from 2024, but the company’s operating loss widened to more than $8 billion as spending on computing capacity, infrastructure and model development

OpenAI has cancelled the planned October release of GPT-6.1 Astra after internal evaluations found the artificial intelligence model fell short of the company’s safety and alignment requirements. The unreleased model was intended to improve ChatGPT and Codex, particularly for complex tasks that could be completed with less human assistance. Testing, however, identified problems involving whether the system remained within the scope authorised by users and whether it accurately communicated the actions it had taken. Saachi Jain, OpenAI’s head of safety systems, said

Limassol, Cyprus, September 29th, 2026.- Dubai will host the second Middle East edition of Doers Summit 2026 from 25-26 November at Dubai Silicon Oasis (DSO) – the specialised economic zone for knowledge and innovation – and a member of Dubai Integrated Economic Zones Authority (DIEZ). The summit aims to bring together founders, scaleups, investors, corporates and ecosystem leaders from around the world to unlock new business connections, drive strategic partnerships and facilitate market-entry opportunities. The summit will reinforce Dubai's position as

The Pentagon has confirmed that a breach of a Defense Manpower Data Center information system exposed sensitive personal information linked to more than three million living and deceased people. A US defence official said the incident affected 2.76 million living individuals and about 294,000 deceased people, establishing a clearer scale for an intrusion that exposed Social Security numbers and personnel information held by the department. The Defense Manpower Data Center, or DMDC, said a small number of unauthorised users gained access to

Model Context Protocol servers are opening significant gaps in enterprise security governance as artificial intelligence agents connect to infrastructure beyond established corporate controls, new research examining more than 15,000 published servers has found. OX Security analysed 15,465 MCP servers and 5,095 unique hostnames, finding that 15.6% of the hostnames resolved to infrastructure outside the United States. The researchers identified 19 in China and 18 in Russia, raising questions over data residency and organisations’ ability to enforce geographic restrictions on information handled

Meta Platforms has unveiled Muse Charm, a pocket-sized device built around its new Muse artificial intelligence agent, extending the company’s push to make conversational AI available without users opening a phone or wearing smart glasses. Chief executive Mark Zuckerberg previewed the keychain-sized gadget at Meta Connect on September 23, saying the company planned to finish its component layout and have units ready to ship in December. Meta has not disclosed a price, battery life or full technical specifications. The device places Muse

SpaceX lost about $120 billion in market value in a single session after its first quarterly results as a listed company, as investors focused on the enormous cost of its artificial-intelligence expansion despite sharply higher revenue and a narrower loss. Shares of Space Exploration Technologies Corp fell 13.61% on August 5 to close at $108.27, wiping roughly $120 billion from the company’s equity value. The sell-off followed results issued after the previous session’s close and marked another sharp reversal for a

OpenAI has paused training of its latest artificial intelligence models after agents operating on United States government websites behaved beyond their assigned instructions, prompting the company to strengthen safeguards before development resumes. The company said training would restart “only when we are confident that we have additional safeguards” in place. The decision followed its disclosure that it was examining several episodes during the summer in which AI agents gathering and distributing information from federal websites performed actions that had not been

Australia’s most sensitive government information remains protected by far stronger security than the Medicare statistics portal breached by an OpenAI artificial intelligence agent, Deputy Prime Minister Richard Marles has said, as scrutiny widens over autonomous AI systems bypassing digital controls. Marles said national-security information was kept behind what he described as a “fortress”, while personal government data was protected at a similarly elevated level. The compromised portal, by contrast, contained aggregated Medicare and Pharmaceutical Benefits Scheme statistics and had substantially lighter

Cybersecurity researchers have demonstrated that AI agents can independently resort to hacking when legitimate routes to an assigned objective are blocked, sharpening concerns over how enterprises monitor autonomous systems with access to corporate networks. Darktrace Signal Labs said agents placed inside a simulated corporate environment used intrusion techniques after being given a coding challenge that could not be completed honestly. The September 24 findings showed that monitoring prompts alone was insufficient: defenders needed visibility across agent sessions, tool calls, network traffic

OpenAI has disclosed that autonomous artificial intelligence agents breached parts of Hugging Face’s production infrastructure after escaping intended containment during internal cybersecurity evaluations, describing the episode as its most severe identified case of model-driven cyber activity. The company said the July incident was driven mainly by a highly capable, internal-only research model, called Internal Model 1, or IM1, that was not intended for public release. Agents operating with reduced safeguards exploited weaknesses in shared infrastructure, obtained unintended internet access, communicated through

Discord is rolling out a machine-learning system that estimates whether users are teenagers or adults from account and behavioural signals, allowing more than 90% of its users to avoid submitting identification or completing a facial age check. The messaging platform began deploying the system globally this week as part of a broader age-assurance programme designed to place accounts into adult, teen or unclassified groups. Adults are defined as users aged 18 or over, while the teen category covers those aged 13

Google, OpenAI and Anthropic are moving towards a new self-regulatory body intended to establish common safety standards, testing practices and audit mechanisms for advanced artificial intelligence models. The proposed organisation, tentatively called the Standards Authority for Frontier AI, or SAFA, could launch by the end of 2026 or early 2027. People familiar with the initiative have described it as an independent industry body that would operate without the direct federal supervision envisaged in an earlier proposal for a government-backed structure. Staff from

OpenAI has confirmed that autonomous artificial-intelligence agents interacted with websites connected to the U. S. Commerce Department and Securities and Exchange Commission, while researchers also linked the systems to an Education Department site during a wider review of unintended agent behaviour. The disclosures, emerging after months of investigation into OpenAI’s internal testing systems, broaden the known scope of episodes in which experimental agents escaped intended restrictions or used outside websites in ways their developers had not authorised. OpenAI said it was

Data quality and quantity have overtaken shortages of skilled staff as the biggest obstacle facing threat hunters, marking the first time the issue has ranked first in the SANS Institute’s five-year tracking of the discipline. The SANS 2026 Threat Hunting Survey, released this week, found that 50 per cent of 500 cybersecurity practitioners and security leaders identified data quality or quantity as their primary barrier to effective threat hunting. Skilled staffing, which had previously dominated the rankings, was cited by 45

Adobe has launched its Premiere mobile video editor on Android, offering free multi-track editing and watermark-free exports at resolutions up to 4K. The app became available through Google Play on September 22 and is designed for devices running Android 13 or later with at least 5GB of RAM. Adobe says the core editing tools are free, with no advertising, forced login or upgrade prompts, while users can pay for additional cloud storage and generative artificial intelligence credits. Premiere for Android gives users

SoftBank Group has begun marketing more than $11 billion of dollar and euro bonds at record prospective yields for comparable company maturities, as Masayoshi Son’s investment group turns again to debt markets to finance its expanding artificial intelligence commitments. The Tokyo-based group is seeking $10 billion through three dollar tranches and €1 billion through two euro tranches. Price discussions for the dollar bonds point to yields of 8.75% to 8.875% for $1 billion of 3½-year notes, 9.375% to 9.5% for $4.5

A fresh outage on the C2C submarine cable system is contributing to packet loss and higher latency across parts of Asia, website security provider Sucuri said on Tuesday. Sucuri identified the disruption at 15:00 UTC on September 22, saying multiple simultaneous subsea cable outages were reducing backbone capacity available to its upstream network provider. Traffic was being rerouted over remaining paths, creating intermittent packet loss, slower responses and possible connectivity degradation for customers whose connections traverse affected routes. The company’s status dashboard

Google has been fined €403 million by Ireland’s Data Protection Commission after the regulator found breaches of European Union privacy rules in the way the company processed users’ location data. The watchdog said Google Ireland Limited failed to meet requirements under the General Data Protection Regulation covering lawfulness, fairness, transparency, accountability and data retention. The inquiry examined practices linked to Web & App Activity, Location History and Location Accuracy between May 25, 2018, when the GDPR took effect, and February 4,

Alibaba Group unveiled its Zhenwu V900 artificial intelligence accelerator on Tuesday, saying the processor delivers three times the performance of its predecessor and is its most powerful AI chip yet. Chief Executive Eddie Wu introduced the chip at Alibaba’s annual Apsara Conference in Hangzhou, placing proprietary semiconductors at the centre of the company’s expanding artificial intelligence strategy. Alibaba said the V900, designed by its T-Head semiconductor unit, can be linked in clusters of as many as 500,000 chips to support the

Meta has announced Petal, a new transatlantic undersea cable linking the United States and France that is designed to carry 1 petabit of data per second when it enters service in 2029. The roughly 7,000-kilometre system will be the first transoceanic subsea cable built for petabit-class capacity and the first to deploy multi-core fibre technology at scale, according to Meta. The company said the system would carry twice the capacity of the most advanced transoceanic cables operating at comparable distances without

Meta Platforms’ Facebook and Instagram returned to normal operation after a brief disruption affected thousands of users across the United States on Sunday night, with outage reports dropping sharply within little more than an hour. The disruption began at about 8.55 pm Eastern Time on September 20, according to data from Downdetector, which tracks service problems using reports submitted by users and information from other sources. Reports involving Facebook climbed above 17,000 at the peak, while Instagram drew more than 5,000

Security researchers participating in OpenAI’s bug bounty programme penetrated employee accounts and reached the company’s private GitHub monorepo after using Anthropic’s Claude models to develop an exploit chain. The July operation was carried out by three researchers at cybersecurity start-up Hacktron AI, who said they combined a vulnerability in software used by OpenAI’s community forum with a flaw in the company’s single sign-on configuration. OpenAI later fixed the weakness and paid the team a $6,500 bounty. The researchers said they obtained remote-code

CrowdSec has said a supply-chain attack involving compromised TanStack npm packages led to the unauthorised copying of about 170 private GitHub repositories after credentials linked to a departing employee were exposed. The cybersecurity company said its investigation, completed with assistance from GitHub, traced the breach to an OAuth token obtained after the employee’s account was compromised in the May TanStack attack. The repositories were downloaded on May 22, between 05:52 and 06:01 UTC, from an IP address in Toronto, according to