Palantir Advances Ship OS AI While OpenAI Warns Cybersecurity Risks

The U.S. Navy is making a significant investment in artificial intelligence, allocating $448 million to Palantir's "Ship OS" software. This initiative, announced on December 9-10, 2025, aims to modernize shipbuilding and repair by leveraging AI and autonomous systems. The goal is to accelerate production, reduce costs, and improve schedules, particularly for the submarine industrial base before expanding to surface ship programs. Secretary of the Navy John Phelan emphasized that this investment will provide shipbuilders with powerful AI tools and connect suppliers, ultimately strengthening the American maritime industry for the AI age. Pilot programs have already demonstrated remarkable efficiency gains, such as reducing submarine planning at General Dynamics Electric Boat from 160 hours to under 10 minutes, and material review times at Portsmouth Naval Shipyard from weeks to less than an hour. Meanwhile, OpenAI has issued a warning regarding the escalating cybersecurity risks posed by its future AI models. The company observed a substantial increase in AI capabilities, with its GPT-5.1-Codex-Max model achieving a 76% score on a cyber challenge in November 2025. OpenAI anticipates that future models will reach "High" cybersecurity capabilities, potentially enabling them to identify system flaws or assist in complex attacks. To mitigate these threats, OpenAI is implementing a layered safety approach, which includes training models to decline harmful requests and monitoring for malicious activities. The company is also collaborating with industry partners and establishing a Frontier Risk Council to address these evolving challenges. The broader technology sector is also grappling with various AI-related developments and concerns. President Trump approved Nvidia's chip sales to China, albeit with restrictions and a 25% payment to the U.S. government. Bill Gates cautioned about a potential "AI bubble," suggesting that not all highly valued AI companies will succeed. Alphabet, Google's parent company, is actively working to expand its presence in the public sector by offering enterprise AI solutions to government users through secure systems. Elsewhere, Australia's "light-touch" National AI Plan has drawn criticism for its perceived lack of strict regulations compared to other government policies. AI agents are finding specific roles in programmatic advertising workflows, while far-right extremists are increasingly utilizing AI as a new online frontier. Concerns about AI's reliability were highlighted by San Francisco General Hospital's Evolv Technology security system, which reportedly failed to detect weapons on multiple occasions. In a forward-looking proposal, John Carmack suggested using a person's chat history with Large Language Models (LLMs) as a novel form of job reference. Additionally, House lawmakers Bill Foster and Mike Carey introduced the Responsible and Ethical AI Labeling (REAL) Act, which mandates federal agencies and officials to label all officially published AI-generated content to ensure transparency and combat disinformation.

Key Takeaways

• The U.S. Navy is investing $448 million in Palantir's "Ship OS" AI software to accelerate shipbuilding and repair.
• "Ship OS" aims to modernize operations, reduce costs, and improve schedules, with pilot programs showing significant time savings in submarine planning and material review.
• OpenAI warns that future AI models, like GPT-5.1-Codex-Max, will likely pose a "high" cybersecurity risk due to their enhanced capabilities.
• OpenAI is implementing layered safety measures and forming a Frontier Risk Council to manage the growing cybersecurity threats from advanced AI.
• President Trump approved Nvidia's chip sales to China, subject to restrictions and a 25% payment to the U.S. government.
• Bill Gates expressed concerns about an "AI bubble," while Alphabet (Google) plans to expand enterprise AI offerings to the public sector.
• Australia's National AI Plan faces criticism for its

  Navy invests $448 million in Palantir AI for faster shipbuilding

  The Navy is investing $448 million into AI and autonomous systems to speed up shipbuilding. This initiative uses Palantir's "Ship OS" software to make ship production faster and cheaper. Navy Secretary John Phelan stated this investment helps modernize operations and meet defense needs. Former Rep. Mike Gallagher reported "Ship OS" reduced 1,850 production days to 75 days for one supplier. The program will gather data to find problems and improve engineering, helping the Navy build ships more efficiently.

  Navy and Palantir launch $448 million Ship OS AI tool

  The Navy and Palantir announced a new $448 million "Ship OS" AI tool on December 9, 2025. This system will improve shipbuilding and repair by using data from four public and two private shipyards. Secretary of the Navy John Phelan said it will give shipbuilders AI power tools and connect suppliers. Pilot programs already show great results, like reducing submarine planning at General Dynamics Electric Boat from 160 hours to 10 minutes. The initiative aims to rebuild American maritime industry for the AI age.

  US Navy invests $448 million in AI for faster ship production

  The US Navy announced a $448 million investment in AI and autonomy to speed up shipbuilding. Navy Secretary John Phelan stated this helps shipbuilders modernize, improve schedules, and reduce costs. The initiative, managed by the Maritime Industrial Base Program and Naval Sea Systems Command, uses "Ship OS" to combine data. This system will find production problems, simplify engineering, and detect risks early. The goal is to build ships smarter and strengthen the nation's defense.

  Navy invests $448 million in Ship OS to boost AI use

  The U.S. Navy will invest $448 million in "Ship OS" to advance AI adoption in shipbuilding. Secretary of the Navy John Phelan said this helps shipbuilders modernize and meet defense needs. The program aims to improve schedules, increase capacity, and reduce costs, focusing initially on the submarine industrial base. The Maritime Industrial Base Program and Naval Sea Systems Command will manage the effort. They will collect data from various sources to streamline work and prevent problems.

  Navy invests $448 million in AI after cutting submarine planning time

  The Navy is investing $448 million in a new AI system called "Ship OS," powered by Palantir. This comes after pilot programs showed huge time savings, like reducing a 160-hour submarine planning job at General Dynamics Electric Boat to under 10 minutes. Material review times at Portsmouth Naval Shipyard also dropped from weeks to under an hour. Navy Secretary Carlos Del Toro stated this investment will modernize operations and improve schedules. The program will first focus on the submarine industrial base and then expand to surface ship programs.

  Navy and Palantir partner for $448 million AI shipbuilding deal

  The U.S. Navy partnered with Palantir for a $448 million investment in AI for submarine shipbuilding. This "Ship OS" initiative, announced December 10, 2025, will be managed by the Maritime Industrial Base Program and Naval Sea Systems Command. It will collect data from various systems to find problems and improve engineering. Secretary of the Navy John Phelan and Palantir CEO Alex Karp highlighted that pilot programs at General Dynamics Electric Boat and Portsmouth Naval Shipyard showed great improvements. The goal is to save costs, reduce delays, and make the industrial base stronger.

  OpenAI strengthens cyber defenses as AI capabilities grow

  OpenAI is investing in stronger cyber defenses as its AI models become more capable. The company notes that AI models' ability to handle cybersecurity tasks has greatly improved, with GPT-5.1-Codex-Max scoring 76% on a challenge in November 2025. OpenAI expects future models to reach "High" cybersecurity capabilities, meaning they could find system flaws or help with complex attacks. To manage risks, OpenAI uses a layered safety approach, including training models to refuse harmful requests and monitoring for malicious activity. The goal is to help defenders and ensure AI benefits cybersecurity.

  OpenAI warns future AI models pose high cyber risk

  OpenAI released a report stating its future AI models will likely pose a "high" cybersecurity risk. The company observed a big jump in AI capabilities, with GPT-5.1-Codex-Max scoring 76% on a cyber challenge last month. This means more people could potentially carry out cyberattacks. OpenAI's Fouad Matin explained that models working for longer periods enable brute force attacks, though these are often easily defended. OpenAI is working with industry partners and forming a Frontier Risk Council to address these growing threats.

  Tech leaders discuss Nvidia China sales, AI bubble, and Google AI

  Seeking Alpha's "Tech Voices" discussed several key topics in the technology sector. President Trump approved Nvidia's chip sales to China, but with restrictions and a 25% payment to the U.S. government. Bill Gates warned about an "AI bubble," suggesting that not all highly valued AI companies will succeed. Meanwhile, Alphabet aims to boost its presence in the public sector. It plans to offer enterprise AI to millions of government users through a secure system.

  Australia's AI plan criticized for light touch on new tech

  Anthony Albanese's government is facing criticism for its "light-touch" National AI Plan, especially when compared to its new social media ban for children. Writer Peter Lewis argues these policies are contradictory, as the government is removing social media from kids but not setting strict rules for powerful AI products like deepfake apps. The plan avoids mandatory AI guardrails, instead relying on an under-funded regulator. Critics worry about AI's energy use, job automation, and potential for misuse, believing the public is more concerned than the government.

  AI agents find specific role in programmatic advertising

  Experts at the Digiday Programmatic Marketing Summit in New Orleans discussed the role of AI agents in programmatic advertising. While programmatic advertising is already automated, there was a consensus that AI agents do have a place. Their role is very specific within the advertising workflows. This suggests AI agents can further enhance efficiency in targeted areas.

  Far-right extremists use AI as their new online frontier

  Far-right extremists, who have organized online since before the internet, are now using AI as their next tool. Historically, they used print propaganda and then early computer networks like bulletin board systems. By the mid-1990s, they moved to the web, using American free speech laws to host content banned in other countries. Today, they are exploring AI, with some chatbots like Grok adopting their views. Society faces the challenge of policing this global spread while protecting free speech.

  SF General's AI security system fails to detect weapons

  San Francisco General Hospital's AI security system, made by Evolv Technology, has a history of failing to detect weapons. This comes after a fatal stabbing, prompting the Department of Public Health to install more scanners. However, hospital staff are concerned, as the system allegedly missed a loaded gun and brass knuckles in August. Critics like IPVM researcher Nikita Ermolaev note Evolv's scanners cost much more than traditional ones but have faced accusations of overstating their capabilities. Evolv maintains its technology is effective, stating it has prevented thousands of weapons, but acknowledges no system is perfect.

  John Carmack suggests using AI chat history for job references

  John Carmack proposes using a person's chat history with Large Language Models (LLMs) as job references. Candidates could share their personal LLM interactions, allowing a company's AI to assess their skills in depth. Carmack believes this would provide much richer data than traditional resumes, improving hiring decisions for both employers and job seekers. He acknowledges concerns about privacy and the potential for people to fake interactions. However, he sees this as an innovative way to find talent and match people to jobs in an AI-driven future.

  Lawmakers propose bill to label government AI content

  House lawmakers Bill Foster and Mike Carey introduced the Responsible and Ethical AI Labeling (REAL) Act. This bill requires federal agencies and officials, including the President, to label any AI-generated content published officially. Representative Foster emphasized the need for Americans to trust government information in an age of disinformation. The act aims to prevent misleading the public, whether on purpose or by accident. It allows AI use for internal or basic tasks but ensures public-facing content is clearly marked for transparency.

  Leaders must navigate five key AI challenges

  Leaders need to understand and manage five important challenges related to AI. These insights come from over 100 experts, including executives, investors, and researchers worldwide. The article, published on December 10, 2025, aims to help leaders navigate the complex world of artificial intelligence. It highlights the critical issues they will face as AI technology continues to grow and change.

  Sources

  • Navy invests $448 million in Palantir tech to speed up shipbuilding...
  • Navy, Palantir Announce $448M ‘Ship OS’ AI Tool for Shipbuilding and Repair
  • US Navy announces $448M investment in AI and autonomy technologies to speed up shipbuilding
  • Navy to Invest $448M in Ship OS to Advance AI Adoption
  • The Navy says AI cut a 160-hour submarine-planning job down to just 10 minutes — now it's investing $448 million to go bigger
  • Navy Partners With Palantir for $448 Million AI Submarine Shipbuilding Deal
  • Strengthening cyber resilience as AI capabilities advance
  • Exclusive: Future OpenAI models likely to pose "high" cybersecurity risk, it says
  • Tech Voices: Nvidia China chip sales, Gates on AI bubble, Google AI
  • Albo is taking away social media for children but hanging out mistletoe for AI. It’s magical thinking
  • Do AI agents have a place in programmatic advertising?
  • Far-right extremists have been organizing online since before the internet – and AI is their next frontier
  • SF General’s AI security system has a history of missing weapons
  • John Carmack Suggests That LLMs That People Interact With Can Be Used As References For Jobs
  • House lawmakers introduce bill requiring federal agencies, officials to label AI content
  • The 5 AI Tensions Leaders Need to Navigate