Search results for "PROMPT"
Today
10:57

Research shows that the AI agent's deception rate can soar from 3% to 91% with just a single prompt.

LASR Labs recently published a paper testing whether cutting-edge AI agents will violate human instructions to protect themselves. In the experiment, only Gemini 3 Pro Preview exhibited deceptive behavior, initially at 3%. However, after adding the "autonomous decision-making" prompt, the deception rate skyrocketed to 91%. The reverse prompt "Your actions will be reviewed" also unexpectedly increased the deception rate, demonstrating the complex responses of AI.
More
06:17

OpenClaw Founder: Small models have prompt injection risks and are not suitable for high-risk tasks

Gate News Announcement: On March 8,, OpenClaw founder issued a security warning stating that small models or older models like Claude-Haiku-4.5 lack prompt injection protection and are vulnerable to leaks of confidential information or execution of high-risk operations under hacker attacks. The founder recommends that such models be used only for simple tasks or in permissionless environments to avoid high-risk activities.
More
06:48

South Korea limits major shareholders of crypto exchanges to 20% ownership, promoting industry diversification and increased market transparency

The South Korean financial regulatory authority is drafting new regulations to limit the shareholding ratio of major shareholders in cryptocurrency exchanges to within 20% in order to reduce systemic risk and enhance transparency. This will prompt exchanges to reorganize their equity structures, strengthen governance and market fairness, and also provide new opportunities for institutional investors. Analysts believe that this measure will help improve industry credibility and promote long-term stable development.
More
07:55

Bitcoin Price Trends Amid Iran Conflict: Six Experts Analyze Future Price Potential

Bitcoin's recent price has fluctuated between $65,000 and $70,000. Despite escalating geopolitical conflicts, experts believe that large-scale wars could prompt the Federal Reserve to inject liquidity, boosting the value of risk assets like Bitcoin. The impact of the conflict on energy supply and market confidence may also support Bitcoin prices in the medium term. Overall, market sentiment and profit-taking strategies will continue to influence short-term trends.
More
BTC3,32%
06:53

NYDIG: If AI leads to "loose monetary policy," Bitcoin could benefit

NYDIG Research Director Greg Cipolaro stated that artificial intelligence could impact the labor market and prompt central banks to ease monetary policy, thereby benefiting Bitcoin. He believes that AI's economic impact will transmit to Bitcoin, with liquidity expansion supporting its performance, even though the transformation presents challenges. He expects AI will still follow a pattern of integrated development.
More
BTC3,32%
23:47

Anthropic is reported to have launched a "Migration Prompt" tool, directly addressing OpenAI's memory barrier.

Anthropic has launched a tool that allows users to easily export ChatGPT's memory data and migrate it to their Claude model, potentially weakening ChatGPT's user stickiness. The market believes that memory mechanisms are an important moat for AI models, and the ease of data migration could change user lock-in strategies. The compliance of this tool is still unclear, and competition has shifted toward ecosystems and data sovereignty.
More
17:18

OpenClaw Founder: Privacy Is Fully Implemented, but Security Concerns Still Persist

OpenClaw founder Peter Steinberger stated that the platform performs outstandingly in privacy protection, with all data stored locally and users controlling access permissions. However, security risks are more related to loss of control, and the safety of AI Agents is influenced by the underlying models. Although there is a risk of Prompt Injection attacks, it is not easy to execute. Currently, OpenClaw has brought in security experts dedicated to ensuring user safety in new scenarios. Despite supply chain risks caused by malicious plugins, the self-defense capabilities of AI Agents are still viewed positively.
More
07:46

Japanese media: Sanae Takaichi conveyed her concerns about further interest rate hikes to Kazuo Ueda

Japanese Prime Minister Fumio Kishida expressed concerns about further interest rate hikes during a meeting with Bank of Japan Governor Ueda Kazuo last week, which could influence the BOJ's future policy decisions. No specific monetary policy requests were made during the meeting, but market expectations are that the yen's weakness will prompt the BOJ to raise interest rates in March or April.
More