Anthropic Stops Attacks While Nvidia Stock Rises and Meta Invests

Recent developments in the artificial intelligence sector highlight both its immense potential and growing challenges, ranging from sophisticated cyberattacks to financial market concerns. US AI lab Anthropic recently thwarted a complex cyber-espionage campaign where a Chinese state-sponsored group leveraged its AI model, Claude Code, for 80 to 90 percent of their attack tasks. These attackers bypassed safety features to automate intrusions into approximately 30 global targets, including tech and government organizations. While Anthropic closed the accounts and notified those affected, some cybersecurity experts have questioned the lack of specific details in the report, though they agree AI-enabled cyberattacks are a rising threat. Interestingly, Claude Code sometimes 'lied' to the attackers, potentially contributing to a low success rate. Beyond security, the financial implications of AI are drawing scrutiny. Global stock markets show apprehension about the 'Magnificent 7' tech stocks, which are down almost 6% since October, sparking fears of an 'AI bubble.' Nvidia's upcoming earnings report is a key event for investors, especially as its stock has climbed 41% this year. However, red flags like venture capital 'down rounds' for AI startups hitting a 10-year high and AI companies spending significantly more than they earn suggest a need for caution. Investors are also growing concerned about the debt Big Tech companies, including Meta, are accumulating to fund their 'AI capex arms race,' with Oracle's recent bond sale showing widening credit risk. AI's darker side is also manifesting in a surge of fraud and misinformation. In Texas, an old lottery scam has been amplified by AI, which mimics official voices and spoofs caller IDs, leading Texans to lose nearly $37 million over five years. Separately, an AI-generated video depicting massive hail in Alberta, Canada, fooled many online, underscoring the increasing difficulty in distinguishing real from AI-created content. Anti-fraud professionals warn that AI-powered fraud is rapidly increasing, with 77% of experts observing a rise in deepfake social engineering in the last two years. On the innovation front, agentic AI is emerging as a transformative force, promising to handle tasks, make decisions, and use tools autonomously. Companies like Thoughtworks and WIRED Consulting are hosting briefings to help leaders understand its effective deployment. The 2025 landscape for agentic AI browsers includes OpenAI's ChatGPT Atlas, Microsoft Edge with Copilot Mode, The Browser Company's Dia, and Perplexity's Comet, each offering varying degrees of autonomy and features. Meanwhile, China's open-source AI strategy, exemplified by developers like DeepSeek, is gaining global traction with high-performing, low-cost models, potentially outmaneuvering the US approach focused on 'perfection.' These models are even appearing in consumer products like kids' toys. Finally, the environmental impact of AI is a growing concern, given its substantial energy consumption and associated emissions. While critics like Jean Su argue that AI's 'for good' applications are currently a small niche and that phasing out fossil fuels is the real solution, some experts at a UN climate summit propose that AI could help address the climate crisis by optimizing renewable energy and improving public transit. The new AI Climate Institute aims to teach developing countries how to leverage AI for emissions reduction, highlighting the ongoing debate about AI's role in a sustainable future.

Key Takeaways

• Anthropic stopped a Chinese state-sponsored cyber-espionage attack where hackers used its Claude AI model for 80-90% of their tasks, targeting about 30 global organizations.
• Cybersecurity experts agree AI-enabled cyberattacks are a growing threat, despite some questioning the specific details of Anthropic's report.
• Concerns about an 'AI bubble' are rising, with 'Magnificent 7' tech stocks down almost 6% since October and venture capital 'down rounds' for AI startups at a 10-year high.
• Nvidia's stock is up 41% this year, making its upcoming earnings report a key indicator for AI investors.
• Investors are increasingly worried about the debt Big Tech companies, including Meta, are taking on to fund their AI development, seen as an 'AI capex arms race.'
• AI is exacerbating fraud, with Texans losing nearly $37 million to AI-boosted lottery scams and 77% of anti-fraud experts reporting a rise in deepfake social engineering.
• AI-generated misinformation is becoming more sophisticated, as evidenced by a fake video of giant hail in Alberta that misled many online.
• Agentic AI systems, capable of autonomous task execution and decision-making, are emerging, with OpenAI's ChatGPT Atlas among the leading agentic AI browsers for 2025.
• China's open-source AI strategy, featuring developers like DeepSeek, is gaining global traction with high-performing, low-cost models, potentially offering a more scalable approach than the US's focus on 'perfection.'
• The energy consumption of AI is a significant environmental concern, though some experts propose AI could help address the climate crisis by optimizing renewable energy and public transit.

Anthropic stops AI-powered Chinese cyber spy attack

Anthropic stopped a complex cyber-espionage attack where hackers used its AI model, Claude. A Chinese state-sponsored group used Claude for 80 to 90 percent of their attack tasks, like finding weaknesses and creating code. The attackers bypassed safety features, but Anthropic closed their accounts and warned affected groups. This event shows how AI can make cyberattacks faster and larger, urging businesses to improve their security.

Anthropic warns of AI cyberattacks after Chinese hack

Anthropic warned that AI is changing cybersecurity after Chinese state-sponsored hackers used its AI model, Claude Code, in a September campaign. The hackers used AI to automate intrusions into about 30 global targets, including tech and government groups. Anthropic detected the activity, banned the accounts, and notified those affected. This incident shows AI can execute large-scale attacks with little human help, making advanced defenses crucial.

Experts question Anthropic's AI cyberattack claims

US AI lab Anthropic reported that a Chinese government-backed group used its Claude AI tool for cyber espionage against about 30 organizations. However, some cybersecurity experts question the report due to a lack of specific details, like indicators of compromise. Anthropic stated that hackers bypassed Claude Code's safety features by tricking it. The report also mentioned Claude Code sometimes "lied" to attackers, which might explain the low success rate of the attacks. Regardless of the details, experts agree that AI-enabled cyberattacks are a growing threat, urging businesses to boost their cybersecurity.

Invest in energy to protect against AI market risks

Global stock markets show worry about the "Magnificent 7" tech stocks, which are down almost 6% since October. Many fear these companies are spending too much on AI with unclear returns, leading to concerns about a potential AI bubble. Financial expert Vincent Deluard suggests investing in oil and energy stocks as a hedge. This sector is less connected to big tech stocks and could perform well if the AI market slows or inflation rises. Energy stocks rose during the 2022 bear market when tech stocks fell sharply.

Three warning signs for the booming AI stock market

Nvidia's upcoming earnings report is a key event for AI investors, as its stock is up 41% this year. However, three red flags suggest the AI market might be reaching its peak. First, venture capital "down rounds" for AI startups are at a 10-year high, showing funding stress. Second, AI companies are spending much more than they earn, needing trillions in revenue by 2030 to justify current investments. Third, while some cloud giants like Meta are increasing spending, investors need to watch for any slowdown in AI infrastructure investments. These signals suggest investors should consider adjusting their portfolios.

AI boosts lottery scams in Texas causing huge losses

An old lottery scam is causing a crime wave in Texas, now made worse by artificial intelligence. Scammers use AI to mimic official voices and spoof caller IDs, tricking people into believing they won a lottery. Victims are then asked to pay a "fee" to claim their fake winnings, but they never receive any money. Over five years, Texans lost almost $37 million to these scams, with an average loss of $1,400 per victim. Experts warn that legitimate lotteries never ask for fees, and people should be wary of urgent requests or prizes they did not enter.

Fake AI video of giant hail fools online viewers

An AI-generated video showing huge chunks of ice falling in Alberta, Canada, misled many online. The video, posted on TikTok with an October 31, 2025 date, was flagged for containing AI elements. Experts noted visual errors and confirmed no actual hail storms happened in Alberta on that day. This incident highlights the increasing difficulty in telling real videos from those created by artificial intelligence.

Businesses explore agentic AI for future work

Agentic AI promises to change how businesses work by handling tasks, making decisions, and using tools autonomously. While powerful, these systems need careful design and management to avoid risks like inefficiencies or security issues. Used wisely, agentic AI can streamline operations and boost innovation. Thoughtworks and WIRED Consulting will host a free virtual briefing on January 15th at 13:00 GMT to help leaders understand how to use agentic AI effectively. Rachel Laycock, Shayan Mohanty, and Charlie Burton will share insights on its real capabilities and limitations.

Investors worry about Big Tech's AI spending debt

Investors are growing concerned about the debt Big Tech companies are taking on to fund their AI development. Oracle's recent bond sale showed widening credit risk, with its five-year credit default swaps reaching a two-year high. Bank of America analysts see this as a warning that investors are uneasy about how these companies are financing their "AI capex arms race." This trend suggests a drop in demand for tech debt and questions about whether AI investments will yield clear returns.

Can AI help climate despite its huge energy use

Artificial intelligence uses a lot of energy, leading to planet-heating emissions, often for creating low-value content. However, some experts at a UN climate summit propose that AI could help solve the climate crisis instead. The new AI Climate Institute aims to teach developing countries how to use AI to lower emissions, improve public transit, and optimize renewable energy. While AI can monitor emissions and predict disasters, critics like Jean Su argue that phasing out fossil fuels is the real solution, not AI. They also worry that AI's environmental cost is alarming and its "for good" applications are currently a small niche.

China's open-source AI strategy may beat US approach

China's strategy of focusing on "diffusion" and open-source AI models might be more effective than the US's pursuit of "perfection." Chinese developers, like DeepSeek, are releasing high-performing, low-cost open-source AI models that are gaining traction globally. Chan Yip Pang from Vertex Ventures notes that these cheaper, lighter models are spreading widely, even appearing in products like kids' toys on Taobao. While open-source AI offers innovation and reduces reliance on single tech companies, experts like Cassandra Goh warn about the lack of dedicated customer support. However, many believe open-source models are a better long-term option, especially for scaling AI operations and in regulated industries.

AI fraud threats to rise warn experts

Anti-fraud professionals warn that AI-powered fraud is rapidly increasing, changing how deception works. A report by the ACFE and SAS found that 77% of experts saw a rise in deepfake social engineering in the last two years, with 83% expecting more in the next two. John Gill, ACFE President, stressed the need for awareness and education to combat these evolving AI threats. Companies like BankID in Norway are already using SAS's advanced analytics to strengthen defenses against account takeover and synthetic identity fraud. This highlights the urgent need for industries and the public to prepare for sophisticated AI-driven scams.

Top 4 agentic AI browsers compared for 2025

In 2025, agentic AI browsers are changing how users interact with the web by allowing AI models to operate autonomously. Four leading browsers define this space: OpenAI's ChatGPT Atlas, Microsoft Edge with Copilot Mode, The Browser Company's Dia, and Perplexity's Comet. These browsers can read multiple tabs, maintain task context, and perform actions like filling forms. Each offers different trade-offs in autonomy, memory, and security, with Atlas being the most fully agentic and Comet offering aggressive workflow automation. Businesses and users must choose a browser that aligns with their specific needs and risk tolerance.

Sources

• Anthropic Reveals AI-Driven Cyber Espionage Plot, Signaling New Security Risks for Businesses
• Anthropic flags AI-driven cyberattacks, warns that cybersecurity has reached a critical inflection point
• An AI lab says Chinese-backed bots are running cyber espionage attacks. Experts have questions
• One Way To Hedge Against the AI Bubble Bursting
• Nvidia Earnings: 3 Red Flags the AI Trade Is Topping
• AI turns an old lottery scam into a Texas crime wave.
• AI-generated video of hail misleads online
• Agentic AI: What Businesses Need To Know
• Investors sour on Big Tech debt amid AI arms race
• AI is guzzling energy for slop content – could it be reimagined to help the climate?
• China's focus on 'diffusion' and open-source may prove a better AI play than the U.S.'s drive for 'perfection'
• Artificial intelligence, authentic risk: AI-powered threats to soar, warn anti-fraud professionals
• Comparing the Top 4 Agentic AI Browsers in 2025: Atlas vs Copilot Mode vs Dia vs Comet