OpenAI and over 100 companies caution that a rise in AI-driven cyberattacks is imminent.

      The coalition asserts that the current security measures will be inadequate as increasingly advanced AI makes sophisticated cyberattacks easier to execute.

      OpenAI, Anthropic, Google, Microsoft, and over 100 other entities are cautioning that AI-driven cyberattacks could become significantly more prevalent and advanced in just a few months.

      The group is advocating for a global initiative to enhance cyber defenses, as more powerful AI models facilitate automation of attacks while complicating containment efforts. They highlight that critical infrastructure like hospitals, water treatment facilities, and internet systems could be at greater risk if defenses do not improve rapidly.

      This warning is particularly significant because it comes from a diverse array of participants within the tech sector. AI creators, cybersecurity firms, and infrastructure providers are all united in the belief that immediate action is necessary from defenders.

      The coalition's wide-ranging warning is noteworthy because its members have differing roles in the cybersecurity landscape. Some are developing powerful AI technologies, while others are focused on defending against cyber threats. This diversity makes the call to action harder to dismiss as merely a single company's agenda. The coalition consists of organizations with varying business interests, yet they collectively agree that there is an urgent need to enhance existing defenses to prevent AI from providing an additional advantage to attackers.

      Their recommendations include the establishment of stronger baseline security standards and broader access to defensive AI solutions, particularly for organizations that may not have substantial cybersecurity budgets. Special attention is given to critical infrastructure, as failures in these areas can rapidly affect multiple entities.

      The coalition seeks to enable governments and organizations to improve access to AI tools that identify vulnerabilities and accelerate patching processes. Smaller organizations are of particular concern due to their potential lack of personnel or resources to respond swiftly to increasingly automated threats.

      Software development also poses a significant challenge. The coalition is advocating for stricter security protocols regarding AI-generated code, as the technology simplifies the process for attackers to identify system vulnerabilities.

      The main approach is to equip defenders with more advanced tools before offensive AI becomes so widespread that it places further strain on already overburdened security teams.

      The urgency of this warning stems from the understanding that AI does not necessarily need to create entirely novel cyberattacks to exacerbate the problem. Merely helping attackers locate existing vulnerabilities more quickly could significantly increase the cost of recognized security failures.

      However, this does not mean that every attack has the potential to become an AI-powered superweapon. Persistent issues like unpatched software and insufficient security personnel still exist, but automation can speed up and simplify the exploitation of these weaknesses.

      The coalition advocates for organizations to fortify their defenses immediately, prior to the common occurrence of AI-driven attacks. For those already struggling to maintain secure and up-to-date systems, delaying actions until attackers enhance their automation capabilities will only leave even less time to catch up.

      South Korea is funding complimentary AI services to assist citizens with appointment scheduling, tax filing, and various daily tasks.

      While global discussions on AI regulation continue, South Korea is instead distributing free AI subscriptions to its citizens. The government aims to make access to artificial intelligence as seamless as electricity or public transport. Rather than requiring costly subscriptions for advanced AI tools, the government plans to cover the expenses, thus offering every citizen complimentary access through a new initiative dubbed 'AI for All'.

      According to the Wall Street Journal, this program intends to promote the use of domestically developed AI chatbots while decreasing the country's reliance on models originating from the United States and China.

      Anthropic has previewed a new standard to facilitate AI-to-machine connections.

      The Model Hardware Standard introduces a universal translation layer, enabling AI agents to have standard controls to operate laboratory instruments and manufacturing machines. Anthropic has rolled out a research preview of its Model Hardware Standard (MHS), a shared specification aimed at allowing AI agents to safely manage lab tools and factory equipment. The company asserts that the MHS dramatically reduces the time necessary for integrating complex machinery with AI from several months to just hours or even minutes. Currently, the standard is accessible to select research facilities and manufacturers via a waitlist, with future plans to make it open-source.

      Gemini Notebook now permits you to reference books and convert them into various formats.

      To use this feature in Gemini Notebook, you must own the books through Google Play Books in order to add them as sources. Gemini Notebook, previously known as NotebookLM, is introducing a new capability that allows users to directly access and cite information from the books they own. This feature, termed Expert Intelligence by Google, aims to clarify the concept behind it. Instead of sourcing information solely from uploaded documents or the internet, users can now gather insights from the digital books they have purchased.

      It is essential to own the book after acquiring it from the Google Play Books library. At present, Google is offering one complimentary book to users in the United States as part of the launch campaign, with a note that this offer is "while supplies last." Currently, Expert Intelligence is available exclusively in the Gemini Notebook application and its web dashboard, although Google has stated that it will