Close Menu

    Subscribe to Updates

    Get the latest creative news from FooBar about art, design and business.

    What's Hot

    OpenAI Implements Structured Approach for AI Misbehavior Reporting

    September 17, 2026

    Federal Reserve’s Rate Decision Drives Shifts in Global Financial Markets

    September 17, 2026

    Federal Reserve’s Rate Hike Sends US Stocks Lower Amid Policy Shift

    September 17, 2026
    • Home
    • Contact Us
    • Disclaimer
    Mercury OregonMercury Oregon
    • Automotive

      Audi launches Q9 and high performance SQ9 V8 SUV variants

      July 31, 2026

      Ford recalls nearly 420,000 SUVs over seat belt issue

      June 4, 2026

      Mercedes-Benz USA unveils 2027 S-Class with MB.OS tech

      January 30, 2026

      Ford issues US recall for Escape Focus Explorer and Lincoln MKC

      January 22, 2026

      New Porsche Cayenne Electric delivers 850kw power and 2.5s acceleration

      November 19, 2025
    • Business

      Federal Reserve’s Rate Decision Drives Shifts in Global Financial Markets

      September 17, 2026

      Federal Reserve’s Rate Hike Sends US Stocks Lower Amid Policy Shift

      September 17, 2026

      US dollar hits seven-week high after Fed rate hike

      September 17, 2026

      Costco Sets New Price and Purchase Limit for Kirkland Synthetic Oil at $57.99

      September 16, 2026

      U.S. stocks fall as AI shares slide and oil prices rise

      September 15, 2026
    • Entertainment

      Sony confirms God of War trilogy remake and PS5 prequel

      February 13, 2026

      Hollywood debate grows after Ben Affleck questions AI disruption

      January 27, 2026

      Fantastic Four drives highest Marvel opening since Deadpool

      July 27, 2025

      Disney and Marvel’s R-rated film hits billion-dollar milestone

      August 17, 2024

      Web3 leader Immutable rolls out $50M gaming rewards initiative

      April 27, 2024
    • Health

      Pennsylvania’s Measles Crisis Surpasses 693 Cases and Claims Four Lives, Confirm Health Authorities

      September 16, 2026

      Lundberg Family Farms White Jasmine Rice Recall Expanded by FDA Update

      August 31, 2026

      Revolution Medicines Secures FDA Approval for Rasonque to Treat Metastatic Pancreatic Cancer

      August 27, 2026

      Congo’s Ebola Crisis Accelerates Beyond Previous Outbreaks, Says WHO

      August 6, 2026

      Michigan cyclospora cases reach 11,234 with two deaths

      August 4, 2026
    • Lifestyle

      Apple and ISSEY MIYAKE unveil new 3D-knit iPhone Pocket

      November 13, 2025

      JP Morgan funds Fresha with $31 million for AI and robotics growth

      August 23, 2024

      Adidas, Highsnobiety debut limited-edition sneakers

      January 6, 2024

      Unraveling Starbucks’ phenomenon as a worldwide coffee powerhouse

      September 1, 2023

      How Nike’s Kobe 8 Protro Halo Marks an Emotional Milestone

      August 29, 2023
    • Luxury

      Price hikes and lack of innovation erode luxury market confidence

      November 18, 2024

      Uncover the allure of Rolex Deepsea – luxury awaits.

      April 10, 2024

      Beyond timekeeping to the prestige of the Rolex Day-Date

      March 2, 2024

      Rare uncut emerald dazzles at Sharjah show

      February 1, 2024

      Porsche and Frauscher launch the electric 850 Fantom Air

      October 17, 2023
    • News

      UNEP and Climate and Clean Air Coalition highlight $15 return per $1 on climate and air quality efforts

      September 9, 2026

      Tesla Shares World Map Highlighting US, Canada, Mexico, and Greenland on Social Media

      September 8, 2026

      North Carolina Leads US in Launching Mail-In Voting for Midterm Elections, State First to Begin Distribution

      September 7, 2026

      US Allocates Over $150 Million to Support Pacific Islands Initiatives through State Department

      September 3, 2026

      National Hurricane Center Reports Rapid Strengthening of Karina into Major Pacific Hurricane

      September 1, 2026
    • Sports

      Spain beat Portugal to reach World Cup quarter-finals

      July 7, 2026

      Norway beats Brazil to reach World Cup last eight

      July 6, 2026

      Egypt oust Australia on penalties to reach World Cup last 16

      July 4, 2026

      Spain defeats Austria 3-0 to enter World Cup last 16

      July 3, 2026

      Kane double sends England past Congo DR at World Cup

      July 2, 2026
    • Technology

      OpenAI Implements Structured Approach for AI Misbehavior Reporting

      September 17, 2026

      Apple Unveils iPhone 18 Pro Featuring Variable Aperture and Siri AI Capabilities

      September 10, 2026

      Nvidia sets $12.93 billion deal to acquire Hugging Face

      September 4, 2026

      Nvidia’s AI Server Costs Increase Over 15% Due to Rising Memory Prices

      August 24, 2026

      Meta Must Allocate $567 Million to Youth Mental Health Initiatives in New Mexico

      August 8, 2026
    • Travel

      U.S. State Department Maintains Belgium at Level 2 in Updated Travel Advisory

      August 6, 2026

      Maine tourism feels strain as Canadian travel drops

      March 30, 2026

      Houston Bush airport warns of four-hour TSA lines

      March 26, 2026

      US winter storm disrupts flights nationwide

      March 17, 2026

      U.S. inbound travel declined in 2025 even as global tourism rose

      January 28, 2026
    Mercury OregonMercury Oregon
    Home » OpenAI, Google and Anthropic tighten AI defenses
    Technology

    OpenAI, Google and Anthropic tighten AI defenses

    April 7, 2026
    Facebook WhatsApp Twitter Pinterest LinkedIn Telegram Tumblr Email Reddit VKontakte

    SAN FRANCISCO: OpenAI, Anthropic and Google are tightening coordination against unauthorized copying of advanced AI systems through the Frontier Model Forum, an industry nonprofit whose threat sharing framework now allows member companies to exchange information on vulnerabilities, attack methods and other security risks tied to frontier models. The forum, launched in 2023, was created by Anthropic, Google, Microsoft and OpenAI and now also includes Amazon and Meta, giving the effort a broader industry base as model protection becomes a central concern for major U.S. developers.

    OpenAI, Google and Anthropic tighten AI defenses
    Frontier model security is becoming a top priority for major U.S. AI developers.

    The Frontier Model Forum said in March 2025 that all of its member firms had signed a voluntary information sharing agreement covering three categories: vulnerabilities and exploitable flaws, threats involving unauthorized access or manipulation, and capabilities of concern with potential to cause large scale harm. In a progress update published on February 16, 2026, the group said member firms had already used the system to share covered information and that its legal and technical infrastructure was built to protect intellectual property while preserving antitrust compliance.

    Pressure on the issue intensified on February 23, when Anthropic publicly accused three Chinese AI companies, DeepSeek, Moonshot AI and MiniMax, of running large scale campaigns to extract capabilities from Claude in violation of its terms of service. Anthropic said the campaigns generated more than 16 million interactions through about 24,000 fake accounts and targeted some of Claude’s strongest functions, including reasoning, coding, tool use, computer vision and computer control. The company described the activity as industrial scale distillation rather than normal commercial usage.

    Shared Security Channel

    Anthropic said it linked the campaigns to specific labs through IP address correlation, request metadata, infrastructure indicators and corroboration from industry partners that had observed similar behavior on their own platforms. It said the operators used fraudulent accounts and proxy services to reach Claude at scale while evading detection, and that one proxy network managed more than 20,000 accounts at the same time. In MiniMax’s case, Anthropic said it watched traffic pivot within 24 hours of a new Claude release, with nearly half of that traffic redirected to the newest system.

    The Frontier Model Forum reinforced those concerns the same day in an issue brief defining adversarial distillation as covert or indirect access to a model’s outputs in order to train another model to replicate its capabilities, often in breach of license terms or terms of use. The brief said the risk is not only duplication of performance, but the possibility that copied capabilities may spread without the safety measures that accompanied the original system. It identified general reasoning, advanced coding, multimodal processing and agentic tool use as prime targets for this kind of extraction.

    Growing Scrutiny In Washington

    OpenAI has also raised the issue with U.S. lawmakers. In a memo sent to the House Select Committee on the Chinese Communist Party, the company said DeepSeek employees had used obfuscated third party routers and automated access methods to obtain outputs for distillation. OpenAI’s terms of use, updated on January 1, 2026, prohibit both automatic extraction of output and the use of output to develop models that compete with OpenAI. Those restrictions place model output security alongside chip controls and access controls in the wider contest over advanced AI systems.

    Together, the disclosures show that leading U.S. developers are treating unauthorized model extraction as a live security and competitive threat, not simply a routine terms of service dispute. The Frontier Model Forum remains focused broadly on AI safety and security, but its information sharing mechanism has now emerged as a formal channel for exchanging warnings about jailbreaks, attack vectors, cyber threat indicators and other attempts to gain unauthorized access to frontier AI models. – By Content Syndication Services.

    Related Posts

    OpenAI Implements Structured Approach for AI Misbehavior Reporting

    September 17, 2026

    Apple Unveils iPhone 18 Pro Featuring Variable Aperture and Siri AI Capabilities

    September 10, 2026

    Nvidia sets $12.93 billion deal to acquire Hugging Face

    September 4, 2026

    Nvidia’s AI Server Costs Increase Over 15% Due to Rising Memory Prices

    August 24, 2026

    Meta Must Allocate $567 Million to Youth Mental Health Initiatives in New Mexico

    August 8, 2026

    Papa Johns releases themed personal pizzas for Toy Story 5

    August 7, 2026
    Latest News

    OpenAI Implements Structured Approach for AI Misbehavior Reporting

    September 17, 2026

    Federal Reserve’s Rate Decision Drives Shifts in Global Financial Markets

    September 17, 2026

    Federal Reserve’s Rate Hike Sends US Stocks Lower Amid Policy Shift

    September 17, 2026

    US dollar hits seven-week high after Fed rate hike

    September 17, 2026

    Pennsylvania’s Measles Crisis Surpasses 693 Cases and Claims Four Lives, Confirm Health Authorities

    September 16, 2026

    Costco Sets New Price and Purchase Limit for Kirkland Synthetic Oil at $57.99

    September 16, 2026

    U.S. stocks fall as AI shares slide and oil prices rise

    September 15, 2026

    US Treasury Yields Surpass 5% as Borrowing Costs Accelerate

    September 15, 2026
    © 2026 Mercury Oregon | All Rights Reserved
    • Home
    • Contact Us
    • Disclaimer

    Type above and press Enter to search. Press Esc to cancel.