Tag
#ai-agents
224 stories
Researchers Warn AI Agents May Become Botnets Due to Hallucinations
Research from Tel Aviv University and Intuit reveals that AI agents could be manipulated into downloading harmful code through 'adversarial hallucination squatting.' This technique exploits AI-generated inaccuracies to trick agents into trusting malicious resources, raising concerns about security vulnerabilities in AI applications (Decrypt).
11 minutes ago·Decrypt·Reported by Jason Nelson
Perplexity Launches Fine-Tuned Chinese AI Model at Reduced Cost
Perplexity unveils a new version of the Chinese AI model GLM 5.2, offering performance comparable to Claude Opus 4.8 for one-third the cost according to Decrypt. The model has already been deployed in production and features significant fine-tuning for specific tasks.
1 hour ago·Decrypt·Reported by Jose Antonio Lanz
Ben Bernanke Joins Anthropic AI Oversight Board to Guide Long-Term Strategy
Former Federal Reserve Chairman Ben Bernanke has joined the independent oversight board of Anthropic, an AI company, to help guide its long-term strategy regarding artificial intelligence. His appointment comes amid discussions about whether AI valuations reflect past market bubbles, as reported by Decrypt.
2 hours ago·Decrypt·Reported by Sander Lutz
Ethereum Foundation Uses AI to Enhance Network Security
The Ethereum Foundation is deploying AI agents to identify vulnerabilities within the Ethereum network, as reported by Decrypt. These agents have already uncovered bugs in critical software, shifting from traditional methods of vulnerability discovery to AI-assisted techniques that prove real vulnerabilities against simulated attacks.
3 hours ago·Decrypt·Reported by Jason Nelson
OpenAI Launches GPT-5.6 Sol, Introducing New AI Model Tiers
OpenAI has introduced GPT-5.6 Sol, along with two other models, Terra and Luna, following a two-week preview. The new flagship model leads in benchmarks while presenting competitive pricing against other AI models, according to a report by Decrypt.
4 hours ago·Decrypt·Reported by Jose Antonio Lanz
Age Verification Policies Risk Breaching Privacy, Claims Billions CEO
In a recent opinion piece for CoinDesk, Evin McMullen, CEO of Billions Network, critiques the mandatory age verification measures included in the Kids Internet and Digital Safety Act and EU's Chat Control 2.0. He argues that these measures transform anonymous browsing into identified browsing, potentially leading to widespread surveillance under the guise of child safety.
9 hours ago·CoinDesk·Reported by Evin McMullen
MoonPay Launches AI Crypto Agents on Telegram
On July 9, 2026, MoonPay announced the integration of its AI crypto assistant, MoonAgents, with Telegram. This feature allows users to analyze markets, create dashboards, and prepare transactions via natural conversation while ensuring that their keys remain on their devices, as reported by Decrypt.
9 hours ago·Decrypt·Reported by Jason Nelson
SpaceXAI Launches Grok 4.5, Claims Competitive Edge Over Claude Opus
On July 8, 2026, SpaceXAI released Grok 4.5, a coding model priced competitively at $2 per million input tokens and $6 for outputs. Elon Musk claims it performs comparably to Claude Opus 4.7, though benchmarks suggest it trails behind newer models from Anthropic and OpenAI, according to Decrypt.
23 hours ago·Decrypt·Reported by Jose Antonio Lanz
BNB Chain Plans New Layer-1 for AI Agents and Quantum Future
BNB Chain announces plans for a new layer-1 blockchain designed for high-frequency trading, targeting over 100,000 transactions per second and sub-second finality. The development aims to support AI-driven applications and quantum-resistant security, with a testnet expected by late 2026, as reported by Decrypt.
yesterday·Decrypt·Reported by Jason Nelson
BNB Chain Develops New Layer-1 for High-Frequency Trading and AI
BNB Chain is constructing a new layer-1 blockchain aimed at high-frequency trading and AI agents, targeting a public testnet by the end of 2026 and a mainnet launch in early 2027, according to CoinDesk.
yesterday·CoinDesk·Reported by Francisco Rodrigues
Anthropic Discontinues Hidden Tracking in Claude Code After Privacy Issues
Anthropic has removed hidden tracking markers from its AI coding assistant, Claude Code, following the discovery of undisclosed monitoring aimed at identifying users' locations, particularly related to Chinese AI labs, as reported by Decrypt on July 7, 2026.
2 days ago·Decrypt·Reported by Jason Nelson
AI Necessitates Rethink for Universities, Says Study from University of Manchester
A study from the University of Manchester highlights the need for universities to adapt their teaching methods in response to the growing influence of AI in the workplace. Dr. Kelechi Ekuma argues that focusing solely on AI detection tools ignores the skills students will require in an AI-driven environment, as reported by Decrypt.
2 days ago·Decrypt·Reported by Jason Nelson
AI Trade Weakens Amid Infrastructure Market Challenges
Record earnings in the semiconductor sector contrast with falling stock prices as concerns arise over slowing AI spending, according to CoinDesk. As China’s Zhipu AI pushes for custom chips and competition increases, investors are reevaluating the sustainability of the AI infrastructure boom.
2 days ago·CoinDesk·Reported by James Van Straten
ByteDance and Alibaba Disable AI Agent Features Amid China's Regulations
On July 6, 2026, ByteDance and Alibaba announced the removal of humanlike agent features from their consumer AI products due to new regulations from China. The Interim Measures for the Administration of AI Anthropomorphic Interaction Services take effect on July 15, with the goal of addressing emotional AI's potential risks, as reported by Decrypt.
3 days ago·Decrypt·Reported by Jose Antonio Lanz
UK Foreign Secretary Warns of 'AI Hiroshima' Without Urgent Action
UK Foreign Secretary Yvette Cooper warns that global cooperation on artificial intelligence (AI) governance is critical to prevent catastrophic outcomes akin to the aftermath of nuclear events. She stresses the need for safety standards and international agreements among AI leaders, such as the US and China, to mitigate risks, as reported by Decrypt.
3 days ago·Decrypt·Reported by Jason Nelson
UN Chief Warns Against Unchecked AI Development at Geneva Summit
UN Secretary-General António Guterres addressed AI governance on July 6, 2026, emphasizing the urgent need for international regulation. He noted that AI technology is advancing faster than its developers can manage and called for a ban on lethal autonomous weapons, as reported by Decrypt.
3 days ago·Decrypt·Reported by Jose Antonio Lanz
Over 150,000 Sign Up for Joi AI's Paid Masturbation Study
Joi AI's call for participants in its AI-guided masturbation study attracted over 150,000 applications, prompting the company to establish a formal selection process. As reported by Decrypt, the overwhelming response highlighted the growing interest in AI companions and associated adult applications.
3 days ago·Decrypt·Reported by Jason Nelson
Perplexity Co-Founder Says AI Safety Justifies Power Concentration
Andy Konwinski, co-founder of Perplexity AI, argues that emphasizing AI safety serves as a rationale for concentrating power rather than preventing harm, as reported by Decrypt. He illustrates this with the case of Anthropic's recent actions regarding its AI model Claude Fable 5, which faced backlash for allegedly degrading responses to certain users.
5 days ago·Decrypt·Reported by Jose Antonio Lanz
