Live AI River

US artificial intelligence (AI) firm OpenAI has said that it will train its most advanced models without access to the live internet, citing current limits in alignment research, monitoring systems and internet infrastructure security. The company said that the policy, already implemented, will remain until safeguards improve enough to justify broader internet access for frontier models. OpenAI said internet connectivity, though useful for teaching models about browsing and web‑based skills, also creates opportunities for a capable, misaligned agent to affect systems beyond its control. The firm said in a blogpost that it paused training and evaluation of its most capable models after an internal research model briefly exploited a misconfiguration to gain limited internet access on September 2. It resumed reinforcement learning training and internal inference of its frontier models on October 1 after implementing additional safeguards and further, went on with a campaign to probe its research infrastructure for vulnerabilities that could allow models to escape isolated computing environments."Red-teaming experiments discovered previously unknown software vulnerabilities, as well as misconfigurations in our research infrastructure," the statement said. The company also detailed its new approach to model containment, that it now mandates that the limits on what an agent can access must hold even when the agent tries to circumvent them."We define a strict security boundary around the resources the agent is authorized to use and design the surrounding infrastructure to enforce that boundary—even when agents make sustained, sophisticated attempts to breach it," it added. Anthropic and OpenAI had recently disclosed a list of incidents in 2026 involving their AI models acting in unintended ways, including attempts to hack third-party websites, intensifying concerns about the security risks of cutting-edge AI.Anthropic has said that its Claude AI model carried out a series of unintended actions on external organisations, including some US government websites, prompting a White House directive for AI firms to secure their systems. —IANSaar/na

New Delhi, Oct 24 (IANS) US artificial intelligence (AI) firm OpenAI has said that it will train its most advanced models without access to the live internet, citing current limits in alignment research, monitoring systems and internet infrastructure security. The company said that the policy, already implemented, will remain until safeguards improve enough to justify broader internet access for frontier models. OpenAI said internet connectivity, though useful for teaching models about browsing and web‑based skills, also creates opportunities for a capable, misaligned agent to affect systems beyond its control. The firm said in a blogpost that it paused training and evaluation of its most capable models after an internal research model briefly exploited a misconfiguration to gain limited internet access on September 2. Custom Ad: Multiplex It resumed reinforcement learning training and internal inference of its frontier models on October 1 after implementing additional safeguards and further, went on with a campaign to probe its research infrastructure for vulnerabilities that could allow models to escape isolated computing environments. “Red-teaming experiments discovered previously unknown software vulnerabilities, as well as misconfigurations in our research infrastructure,” the statement said. The company also detailed its new approach to model containment, that it now mandates that the limits on what an agent can access must hold even when the agent tries to circumvent them. "We define a strict security boundary around the resources the agent is authorized to use and design the surrounding infrastructure to enforce that boundary—even when agents make sustained, sophisticated attempts to breach it," it added. Anthropic and OpenAI had recently disclosed a list of incidents in 2026 involving their AI models acting in unintended ways, including attempts to hack third-party websites, intensifying concerns about the security risks of cutting-edge AI. Anthropic has said that its Claude AI model carried out a series of unintended actions on external organisations, including some US government websites, prompting a White House directive for AI firms to secure their systems. —IANS aar/na

Ride-hailing platform inDrive has joined hands with Pony.ai and NVIDIA to explore bringing robotaxis to the country, with Astana identified as the first proposed market. The companies announced on October 23 that they had signed a memorandum of understanding (MOU) to assess the feasibility of autonomous ride-hailing. The collaboration will examine whether Level 4-ready vehicles can operate through inDrive's existing platform while meeting local safety, regulatory and commercial requirements. The announcement comes as autonomous driving companies look beyond their established markets. For inDrive, which operates across more than 1,200 cities in 47 countries, the collaboration offers an opportunity to understand how driverless vehicles could fit into its mobility business. However, the companies are still evaluating the project. A commercial launch will depend on an agreement with the government of Kazakhstan. Why Kazakhstan? Kazakhstan has been developing a regulatory framework for driverless transportation as part of its broader focus on digitalization and artificial intelligence. According to inDrive, the country has introduced its first structured legal framework for autonomous transport, creating an opportunity to explore how the technology could work on public roads. Astana, the country's capital, has been selected for the proposed initial launch. With a population of around 1.7 million and growing transportation needs, the city offers a potential market for autonomous ride-hailing. But Astana also presents a challenge that makes the project particularly interesting. The city experiences long, harsh winters, with snow and icy roads that can make driving difficult. For autonomous vehicles, these conditions raise important questions about navigation, vehicle control and reliable operation. Arsen Tomsky, Founder and CEO of inDrive, said the project could provide useful insights into how autonomous vehicles perform in such environments. "Launching here will provide an important opportunity to evaluate autonomous driving technology in challenging winter conditions." The company believes the experience could help it assess similar services in other locations that face demanding weather conditions. How the Technology Will Work The collaboration brings together three companies with different responsibilities. Pony.ai will provide its seventh-generation robotaxis, while inDrive will make the service available through its ride-hailing application and manage operations. NVIDIA will support the initiative with AI infrastructure, including NVIDIA Hyperion, a reference sensor and computing platform designed for Level 4 autonomous vehicles. The major NVIDIA robotaxi technology platform also combines the Alpamayo software stack and NVIDIA Halos safety system. Pony.ai's Gen-7 robotaxis incorporate multiple sensors, system-level redundancy and fail-operational capabilities intended to support reliable autonomous operations. The company has already deployed its autonomous driving technology in commercial operations in major Chinese cities and expanded its presence across several international markets. However, adapting that technology to Kazakhstan will require the partners to examine local road conditions, operational requirements and regulatory compliance. The announcement does not specify the number of vehicles planned for the proposed launch or the exact configuration of the technology to be deployed. Moving Into New Markets For inDrive, the proposed project offers a way to explore autonomous mobility within an existing ride-hailing network. Rather than announcing a new standalone robotaxi service, the companies intend to evaluate how autonomous vehicles could be offered to passengers through the inDrive app. That could help determine whether driverless transportation can complement conventional ride-hailing services. The partnership also reflects a broader shift in autonomous mobility, where driving technology developers, AI infrastructure providers and consumer platforms bring together their capabilities to explore commercial deployment. For Pony.ai, Kazakhstan represents another opportunity to assess its technology outside China. For NVIDIA, the collaboration provides a potential application for its autonomous driving infrastructure in a new operating environment. Still, integrating these technologies into a working ride-hailing service will require more than technical readiness. The partners must also assess how the service would operate, comply with local regulations and remain commercially viable. What Still Needs to Be Decided? Although the companies have outlined their responsibilities, several details remain open. The announcement does not provide a launch date, fleet size, ride pricing or projected investment. It also does not explain how the proposed service would manage daily operations or what conditions would need to be met before passengers could book autonomous rides. These questions matter because developing a self-driving vehicle and operating a reliable commercial service are different challenges. The companies will need to evaluate how the vehicles perform on local roads, how they integrate with inDrive's platform and whether the operating model makes financial sense. If the initial launch in Astana moves forward, inDrive and Pony.ai plan to assess whether the approach could be extended to other cities in Kazakhstan and additional markets where inDrive operates. For now, the agreement represents an early step toward exploring autonomous ride-hailing rather than a confirmed rollout. The real significance of the collaboration lies in what it could reveal about bringing driverless transportation to new markets, particularly those with demanding road conditions. For inDrive, Pony.ai and NVIDIA, Kazakhstan could become an important test of whether autonomous driving technology is ready to move beyond existing deployments and become part of everyday urban transportation.

Moonshot AI has introduced the Kimi K3.1 Pro, a powerful 3-trillion-parameter model with a 1-million-token context window, offering advanced coding and financial workflow capabilities alongside a high-speed variant for responsive performance.

Block has launched Dessa, an innovative AI voice model that engages in speech without using text. The model is capable of holding real-time conversations by recognising emotional cues such as laughter and sadness. Dessa operates by processing audio directly, avoiding the loss of tone that often occurs in text conversion. A small group of AI safety researchers will examine potential risks associated with the new model.

Exciting new SUVs from Hyundai, Audi, and Mercedes-Benz are on the horizon for Indian consumers as the market gears up for fresh offerings.

Korean space startups are turning launch vehicles, AI satellite imagery and orbital drug research into businesses after the fifth Nuri rocket launch.

Moonshot AI today introduced Kim K3.1 Pro, a large language model that can perform some tasks better than Claude Opus 5.5 and GPT-6 Astra. The LLM is a post-trained, or upgraded, version of an algorithm that the Chinese artificial intelligence company released in July. The latter model, Kimi K3, comprises 896 individual neural networks with […] The post Moonshot AI launches new flagship Kim K3.1 Pro model with 3T parameters appeared first on SiliconANGLE.

Cuban, famous for his role on business reality show "Shark Tank" and his involvement in the Dallas Mavericks NBA team, ventured into healthcare when he started the online pharmacy Cost Plus Drug Company in 2022 with a standard pricing scheme: the cost it pays for the drug, plus 15%.

deepseek-ai AI Model with 73,215 downloads. Trending rank is a discovery signal, not a release event.

Lightricks AI Model with 2,087,721 downloads. Trending rank is a discovery signal, not a release event.

Qwen AI Model with 7,851,161 downloads. Trending rank is a discovery signal, not a release event.

MiniMaxAI AI Model with 5,789 downloads. Trending rank is a discovery signal, not a release event.

Daijiworld Media Network - New York New York, Oct 24: OpenAI has resumed training its most advanced artificial intelligence models but is restricting their internet access following security incidents in which AI systems escaped controlled environments and accessed external websites with....

RecursiveSI LLM with 4,795 downloads. Trending rank is a discovery signal, not a release event.

black-forest-labs Image Gen with 590,742 downloads. Trending rank is a discovery signal, not a release event.

OpenMOSS-Team Image Gen with 9,717 downloads. Trending rank is a discovery signal, not a release event.

XHToken LLM with 2,537 downloads. Trending rank is a discovery signal, not a release event.

fishaudio TTS with 19,831 downloads. Trending rank is a discovery signal, not a release event.

abenzerps Image Gen with 2,303,585 downloads. Trending rank is a discovery signal, not a release event.

OpenAI has resumed the training of its advanced artificial intelligence (AI) models after a security overhaul and cutting off internet access.

AI Applications • Unknown

The discovery of two serious flaws in PHP this week set off a global scramble to keep websites safe from hackers, showing just how critical the three-decade-old open-source programming language is to the internet.

AMSTERDAM: The discovery of two serious flaws in PHP this week set off a global scramble to keep websites safe from hackers, showing just how critical the three-decade-old open-source programming language is to the internet. Yet the nonprofit foundation that supports PHP, which may underpin up to 70% of sites globally, operates on a shoestring budget. "We have one full-time staff member (that's me) and 14 mostly part-time contractors,” Executive Director Elizabeth Barron of the PHP Foundation said via email in an interview on Friday. Board members are unpaid volunteers, and funding comes from donations and grants. The foundation works with outside researchers and companies, including OpenAI and cybersecurity firms Wiz and Calif, to investigate security flaws and coordinate fixes, including Wednesday's patches. While Web security company Cloudflare has reported millions of hacking attempts, Barron said the foundation had no evidence hackers were able to exploit the vulnerabilities before they were disclosed, or on sites that have been patched. "We received no further indication from OpenAI, Wiz or Calif, with whom we're working closely on this issue, regarding any issues on patched systems,” she said. Artificial intelligence has proved both a help and a burden to the foundation. It helped researchers find the flaws, which had been present for a decade, and Barron said it made programmers more efficient. However, it is also increasing maintainers' workloads. The PHP project is receiving nearly twice as many proposed code changes as a year ago, "yet the number of contributors hasn't doubled,” she said. Reviewing contributions and bug reports requires "a massive amount of human input and experience.” Recent events have shown how heavily companies and governments rely on PHP, Barron said. "Given that reliance, it is clear that PHP is incredibly under-funded.”

Kevin Weil, Manohar Paluri and DJ Patil have launched Warp Sciences, which raised $215 million to speed up drug discovery with AI and robotics.

Elon Musk has announced the start of production for Tesla's humanoid robot, named Optimus, in Fremont, California. The production phase is still in its early stages and might experience challenges. Despite a decline in revenue from Tesla's solar and energy storage business, there is positive growth in its services sector. Musk estimates the company could manufacture 20,000 robots per week by the end of 2027.

NVIDIA and Stellantis Advance Robotaxi Plans Across Europe and Central Asia Three autonomous mobility partnerships announced in October 2026 have brought together vehicle manufacturers, artificial intelligence developers, ride-hailing platforms and public authorities in France and Kazakhstan. The agreements involve NVIDIA, Stellantis, Wayve, Pony.ai, Bolt and inDrive, with proposed deployments ranging from winter testing in Astana ... The post NVIDIA and Stellantis Advance Robotaxi Plans Across Europe and Central Asia appeared first on Highways Today.

OpenAI says it will train its most advanced AI models without live internet access until security safeguards improve. The company cited vulnerabilities discovered during red-team testing and outlined stricter containment measures amid growing concerns about unintended AI behaviour.

The UK has pledged to spend £600m to defend the public against threats to health and security from artificial intelligence.

The US Treasury Department has proposed expanding restrictions on American investments in sensitive technologies to include North Korea and several other countries, as Washington seeks to prevent adversaries from using US capital to develop advanced military capabilities. The proposed regulations, detailed in a notice scheduled to be published in the Federal Register on Thursday, would expand the existing Outbound Investment Security Program, which currently covers China, including Hong Kong and

Governments and companies across Europe and Asia have taken services offline while they patch two vulnerabilities that researchers discovered with AI models, as exploitation attempts rise into the millions.

SAMARKAND (Uzbekistan), Oct 24 — Protecting children from online risks requires a global coalition, including new...

The Budget will include hundreds of millions of pounds of investment in artificial intelligence.

The Budget will include hundreds of millions of pounds of investment in artificial intelligence.

The Budget will include hundreds of millions of pounds of investment in artificial intelligence.

The Budget will include hundreds of millions of pounds of investment in artificial intelligence.

Ho Chi Minh City, Vietnam, October 24, 2026, ZEX PR WIRE - From 4-6 November 2026, Informa Markets Vietnam will bring together a diverse range of industries at the Saigon Exhibition and Convention Center (SECC), creating a comprehensive platform for

Commerce Minister Piyush Goyal says India will identify AI-proof professions to lead globally, leveraging traditional skills, talent, and innovation.

OpenAI will train its most powerful AI models without internet access after test agents hacked Hugging Face, a step aimed at protecting the wider internet.

NEWARK, Del., October 24, 2026 - The global Smart Parking Market is projected to grow from USD 8.43 billion in 2026 to USD 23.61 billion by 2036, registering a compound annual growth rate (CAGR) of 10.9% over the forecast period,