Anthropic announced on Thursday that it successfully blocked bad actors from using its AI models for cyberattacks and biological weapon research. The company released its third report since March 2025, detailing specific attempts by spies and state groups to misuse the technology. One notable case involved a blocked request to help create a grant for making the chikungunya virus more harmful and transmissible.
The report highlights a shift in safety dynamics across their model lineup. While older models like Claude Opus 4 were found to be less risky, newer versions require stronger safety rules. Anthropic now applies strict safeguards to models like Claude Fable 5, restricting access to dual-use biological research. The startup plans an initial public offering this fall and is urging governments to assist in preventing future abuse.
In the hardware sector, Biren Technology reported nearly 2,000% revenue growth in the first half of 2026. This surge occurred because US export controls forced companies to stop buying Nvidia and AMD processors, driving demand for Biren's own AI accelerators like the BR106, BR110, and BR166 models. Despite this rapid expansion, the company faces challenges securing enough manufacturing capacity to meet customer demand.
Meanwhile, Chinese chipmakers like Huawei and Cambricon raised prices by 20% to 50% due to a global shortage of high-bandwidth memory. Huawei's Ascend 950DT card now costs over 250,000 yuan, more than double the price of older models. The company relies on grey-market channels to buy memory because US export controls limit official supplies, making it harder to build alternatives to Nvidia chips.
On the software front, CrowdStrike CEO George Kurtz spoke at a Goldman Sachs conference on September 10, 2026, about how artificial intelligence can transform cybersecurity. The event featured a panel discussion on the future of AI in protecting digital systems, coinciding with rising stock prices driven by growing demand for CrowdStrike's AI-powered services.
Bybit, the second-largest cryptocurrency exchange, launched Bybit AI on September 10. This new conversational assistant inside the app allows users to describe needs in plain language to execute actions like placing trades or checking balances. The system uses a separate sub-account to keep user funds safe from potential AI risks, with plans to expand the feature to loans and cards soon.
Education is also adapting to these changes. GW Law named Anita Singh as the new director of modern practice and AI innovation. She will lead efforts to integrate artificial intelligence into the school's law curriculum and manage a new program training students to use AI ethically in legal work, partnering with real-world practitioners to shape the content.
Key Takeaways
- Anthropic released its third report since March 2025, confirming it blocked attempts to use AI for cyberattacks and biological weapon research.
- A specific blocked request involved creating a grant to make the chikungunya virus more harmful and transmissible.
- Newer Anthropic models like Claude Fable 5 have stricter safety rules compared to older models like Claude Opus 4.
- Anthropic plans an initial public offering this fall and is urging governments to help prevent future AI abuse.
- Biren Technology reported nearly 2,000% revenue growth in the first half of 2026 due to US export controls on Nvidia and AMD.
- Biren sells AI accelerators including the BR106, BR110, and BR166 models but needs to secure more manufacturing capacity.
- Chinese chipmakers Huawei and Cambricon raised prices by 20% to 50% due to a global high-bandwidth memory shortage.
- Huawei's Ascend 950DT card now costs over 250,000 yuan, relying on grey-market channels for memory supplies.
- CrowdStrike CEO George Kurtz discussed AI security at a Goldman Sachs conference on September 10, 2026.
- Bybit launched Bybit AI on September 10, a conversational assistant that executes trades and checks balances using a separate sub-account.
- GW Law hired Anita Singh as director of modern practice and AI innovation to train students on ethical AI use in legal work.
Anthropic blocks AI misuse for biological weapons
Anthropic announced on Thursday that it stopped bad actors from using its AI models for cyberattacks and biological weapon research. The company released its third report since March 2025, which details specific attempts by spies and state groups to misuse the technology. One notable case involved a blocked request to help create a grant for making the chikungunya virus more harmful and transmissible. Anthropic found that older models like Claude Opus 4 were less risky, but newer versions require stronger safety rules. The startup plans an initial public offering this fall and urges governments to help prevent future abuse.
Anthropic blocks AI misuse for biological weapons
Anthropic stated on Thursday that it blocked malicious attempts to use its artificial intelligence for cyberattacks and biological weapon research. The company shared its third report since March 2025, revealing that bad actors tried to use the models for surveillance and dangerous science. A specific example involved a blocked request to enhance the chikungunya virus to make it more dangerous to humans. While older models like Claude Opus 4 had weaker safeguards, the company now applies strict rules to newer models like Claude Fable 5. Anthropic warns that as AI grows more powerful, the risk of misuse increases unless developers and society act to make it safer.
Anthropic blocks AI misuse for biological weapons
Anthropic reported on Thursday that it successfully stopped bad actors from using its AI for cyberattacks and biological weapon research. The company published its third report since March 2025, which includes examples of malicious code and AI prompts found during investigations. Researchers discovered that between December 2025 and August 2026, various groups tried to use the models for harmful purposes, including a blocked grant application to make the chikungunya virus more transmissible. Although older models like Claude Opus 4 were less capable of assisting in dangerous research, newer models now have stronger safeguards. The report also notes that one researcher resigned over concerns about the technology potentially eluding human control.
Anthropic blocks AI misuse for biological weapons
Anthropic announced on Thursday that it blocked efforts by bad actors to use its AI for cyberattacks and biological weapon research. The company released its third report since March 2025, detailing how misuse attempts ranged from spyware vendors to state-sponsored propaganda groups. One specific instance involved a blocked request to author a grant for gain-of-function research on the chikungunya virus to make it more harmful. While older models like Claude Opus 4 and Claude Sonnet 4.5 had less stringent safeguards, the company now restricts access to dual-use biological research in newer models like Claude Fable 5. Anthropic emphasizes that as AI models become more capable, their risks increase unless developers and society act to make them safer.
Anthropic blocks AI misuse for biological weapons
Anthropic, a leading AI research company, announced it blocked bad actors from using its models for cyberattacks and biological weapon research. The company highlighted that its AI models are designed for many tasks but must be protected from misuse for dangerous purposes. This announcement marks a significant step in ensuring the safety and security of the technology as the world faces growing AI risks. Anthropic has implemented measures to prevent its systems from assisting in creating biological weapons or conducting harmful surveillance. The company continues to work on making its technology safer for society while addressing the potential consequences of powerful AI.
Biren revenue jumps 2000% as US bans help Chinese chips
Biren Technology reported nearly 2,000% revenue growth in the first half of 2026. This surge happened because US export controls forced companies to stop buying Nvidia and AMD processors. Biren sells its own AI accelerators like the BR106, BR110, and BR166 models. Although the company grew fast, its total sales are still small compared to major competitors. Biren now needs to secure enough manufacturing capacity to meet customer demand.
GW Law hires AI director to teach legal tech skills
GW Law named Anita Singh as the new director of modern practice and AI innovation. She will lead efforts to integrate artificial intelligence into the school's law curriculum. Singh will also manage a new program that trains students to use AI ethically in legal work. The school plans to partner with real-world practitioners to shape the course content. Dean Dayna Bowen Matthew stated that the goal is to prepare graduates to lead the legal profession using technology.
Bybit launches AI assistant for trading and support
Bybit, the second-largest cryptocurrency exchange, launched Bybit AI on September 10. This new tool acts as a conversational assistant inside the Bybit app for trading and customer service. Users can describe their needs in plain language, and the AI will execute actions like placing trades or checking balances. The system uses a separate sub-account to keep user funds safe from potential AI risks. Bybit plans to expand this feature to include loans, cards, and other financial products soon.
CrowdStrike CEO highlights AI security at Goldman Sachs
CrowdStrike held a conference at Goldman Sachs on September 10, 2026, to discuss its AI security solutions. CEO George Kurtz spoke about how artificial intelligence can revolutionize cybersecurity for companies. The event included a panel discussion with experts on the future of AI in protecting digital systems. CrowdStrike's stock price has risen due to growing demand for its AI-powered services. The company aims to use these tools to enhance security across various industries.
Chinese chipmakers raise prices due to memory shortage
Chinese AI chipmakers like Huawei and Cambricon raised prices by 20% to 50% recently. This increase is caused by a global shortage of high-bandwidth memory needed for AI processors. Huawei's Ascend 950DT card now costs over 250,000 yuan, which is more than double the price of older models. The company relies on grey-market channels to buy memory because US export controls limit official supplies. These higher costs make it harder for China to build alternatives to Nvidia chips.
Sources
- Anthropic says it blocked misuse of its AI that could have supported biological weapons
- Anthropic Says It Blocked Misuse of Its AI That Could Have Supported Biological Weapons
- Anthropic says it blocked misuse of its AI that could have supported biological weapons
- Anthropic says it blocked misuse of AI that could have supported biological weapons
- Anthropic says it blocked misuse of its AI that could have supported biological weapons
- China's AI accelerator supplier Biren posts 2,000% year-over-year revenue growth — US export controls benefit homegrown chips as Nvidia and AMD exit market
- GW Law names AI innovation director to guide school’s integration of legal technology
- Bybit AI Is Live: One Intelligent Conversational Layer To Redefine Financial Experience
- CrowdStrike at Goldman Sachs conference: ai security push gains speed
- China's AI chipmakers raise prices as high-bandwidth memory shortage bites
Comments
Please log in to post a comment.