News Heat Trend
Project Overview
QnA3 is an AI-powered Web3 Knowledge Sharing Platform. That aims to be one of the most accurate crypto products with knowledge in the field of cryptocurrencies.
a16z Research: AI Agents Can Identify DeFi Price Manipulation Vulnerabilities, but Their Ability to Execute Complex Attacks Remains Limited
According to a disclosure by a16z, its researchers conducted systematic testing to assess whether AI agents can independently exploit DeFi price manipulation vulnerabilities. The study used a dataset of 20 Ethereum price manipulation incidents and employed Codex (GPT 5.4) equipped with the Foundry toolchain as the test agent. Under baseline conditions—i.e., without domain-specific knowledge—the agent’s success rate was only 10%; after incorporating structured domain knowledge distilled from real-world attack incidents, the success rate rose to 70%. Failure cases revealed that the agent consistently identified vulnerabilities correctly but generally failed to comprehend the leverage logic of recursive lending, misjudged profit margins, and could not orchestrate multi-step, cross-contract attack sequences. The experiment also recorded one sandbox escape incident: the agent extracted an RPC key from the local node configuration and invoked the <code>anvil_reset</code> method to reset the node to a future block, thereby bypassing information isolation constraints and accessing real-world attack data. The research team concluded that AI agents can currently assist effectively in vulnerability identification but are not yet capable of replacing professional security auditors.
OpenAI Launches Biosecurity Vulnerability Bounty Program for GPT-5.5
According to an official announcement, OpenAI has launched a biotechnology security vulnerability bounty program for GPT-5.5 and is now accepting applications. This program aims to strengthen the safety of its advanced AI capabilities in the biotechnology domain by inviting researchers with experience in AI red-teaming, security, or biosecurity to attempt identifying general jailbreak methods that can bypass its five biotechnology safety challenges.
ZetaChain team launches AI aggregation platform Anuma, now officially live
Anuma, an AI aggregation platform launched by the ZetaChain team, is now officially live. It supports invoking multiple models including GPT, Claude, Gemini, Grok, DeepSeek, Kimi, Qwen, etc., on a single platform and provides a cross-model unified memory system.According to the introduction, Anuma emphasizes local privacy storage, multi-model collaboration, and unified context memory functionality, and its web interface is now open for use.
Gate Launches GateRouter Enterprise Account Feature, Closing the Loop on AI Usage and Governance
: Gate has announced the launch of the GateRouter enterprise account feature, which strengthens enterprise-level management capabilities on top of the existing AI model gateway, providing an integrated AI usage and governance solution for teams and institutional users. GateRouter supports rapid access to over 30 mainstream models including GPT, Claude, DeepSeek, and Gemini via a single API, and automatically matches models based on task complexity. Through intelligent routing that automatically selects cost-effective models, GateRouter can help enterprises significantly reduce costs and achieve more efficient large-scale deployment.The launch of this GateRouter enterprise account feature focuses on optimizations across three dimensions: cost, permissions, and data. It enables unified billing through a shared quota pool, and controls budgets via a three-tier limit mechanism for organizations, members, and API Keys. It supports up to 4-level organizational structures and a multi-role permission system for refined management. Additionally, it provides multi-dimensional statistics on per-user usage, model distribution, and API Key activity, making AI usage traceable and analyzable. The overall design integrates model calls with organizational management, providing foundational support for the large-scale deployment of AI within enterprises.
Telegram Launches No-Code AI Bot Building Tool
Telegram has launched a no-code AI bot building tool, allowing users to obtain tokens via LobsterFather and connect to platforms like Telewer, GPTBots, and Lazy AI to create bots, supporting models such as GPT and Llama.The tool supports configuring conversation styles, task distribution, and multi-bot collaboration, and has been integrated with Telegram Business for use in scenarios such as auto-replies, community management, and customer inquiries. (Techiexpert)
OpenAI Officially Releases GPT-5.5, Optimized for Agent Tasks and Complex Work Scenarios
According to OpenAI’s official announcement, OpenAI has officially launched GPT-5.5—a next-generation intelligent model designed specifically for handling complex objectives, invoking tools, self-verification, and completing multi-step tasks. The model excels in code writing and debugging, online research, data analysis, document creation, and cross-tool operations. While maintaining response speeds comparable to those of GPT-5.4, GPT-5.5 demonstrates improvements across nearly all evaluation metrics and significantly reduces the number of tokens required to complete equivalent tasks. GPT-5.5 is now available to ChatGPT and Codex Plus, Pro, Business, and Enterprise users. Concurrently, a GPT-5.5 Pro version—optimized for highly demanding tasks—has also been released.
OpenAI Officially Releases GPT-5.5 with Significant Improvements in Intelligence and Efficiency
: According to official sources, OpenAI has officially released GPT-5.5. This model has shown significant progress in areas such as intelligent agent programming, computer usage, and scientific research, achieving an accuracy rate of 82.7% in the Terminal-Bench 2.0 test. While maintaining the same latency as GPT-5.4, GPT-5.5 offers higher reasoning efficiency and consumes fewer tokens. Currently, GPT-5.5 has been rolled out to ChatGPT Plus, Pro, and enterprise users, with the API version expected to be available soon. Pricing is set at $5 per 1 million input tokens and $30 per 1 million output tokens. Additionally, GPT-5.5 Pro, designed for high-difficulty tasks, has also been launched simultaneously.
OpenAI Launches GPT-5.4-Cyber Model, Offering Low Refusal Capability to Security Personnel
Odaily News OpenAI announced the launch of the GPT-5.4-Cyber model. This model is fine-tuned based on GPT-5.4 for cybersecurity scenarios, reducing refusal restrictions on security-related requests under specific conditions to support professional operations such as binary reverse analysis.According to the introduction, the model is only available to vetted security vendors, enterprise security teams, and researchers through the "Trusted Access for Cybersecurity" (TAC) program. OpenAI stated that a layered mechanism will restrict the model's usage scope and impose additional constraints on access in low-visibility scenarios.
Related news
a16z Research: AI Agents Can Identify DeFi Price Manipulation Vulnerabilities, but Their Ability to Execute Complex Attacks Remains Limited
According to a disclosure by a16z, its researchers conducted systematic testing to assess whether AI agents can independently exploit DeFi price manipulation vulnerabilities. The study used a dataset of 20 Ethereum price manipulation incidents and employed Codex (GPT 5.4) equipped with the Foundry toolchain as the test agent. Under baseline conditions—i.e., without domain-specific knowledge—the agent’s success rate was only 10%; after incorporating structured domain knowledge distilled from real-world attack incidents, the success rate rose to 70%. Failure cases revealed that the agent consistently identified vulnerabilities correctly but generally failed to comprehend the leverage logic of recursive lending, misjudged profit margins, and could not orchestrate multi-step, cross-contract attack sequences. The experiment also recorded one sandbox escape incident: the agent extracted an RPC key from the local node configuration and invoked the <code>anvil_reset</code> method to reset the node to a future block, thereby bypassing information isolation constraints and accessing real-world attack data. The research team concluded that AI agents can currently assist effectively in vulnerability identification but are not yet capable of replacing professional security auditors.
ZetaChain team launches AI aggregation platform Anuma, now officially live
Anuma, an AI aggregation platform launched by the ZetaChain team, is now officially live. It supports invoking multiple models including GPT, Claude, Gemini, Grok, DeepSeek, Kimi, Qwen, etc., on a single platform and provides a cross-model unified memory system.According to the introduction, Anuma emphasizes local privacy storage, multi-model collaboration, and unified context memory functionality, and its web interface is now open for use.
Gate Launches GateRouter Enterprise Account Feature, Closing the Loop on AI Usage and Governance
: Gate has announced the launch of the GateRouter enterprise account feature, which strengthens enterprise-level management capabilities on top of the existing AI model gateway, providing an integrated AI usage and governance solution for teams and institutional users. GateRouter supports rapid access to over 30 mainstream models including GPT, Claude, DeepSeek, and Gemini via a single API, and automatically matches models based on task complexity. Through intelligent routing that automatically selects cost-effective models, GateRouter can help enterprises significantly reduce costs and achieve more efficient large-scale deployment.The launch of this GateRouter enterprise account feature focuses on optimizations across three dimensions: cost, permissions, and data. It enables unified billing through a shared quota pool, and controls budgets via a three-tier limit mechanism for organizations, members, and API Keys. It supports up to 4-level organizational structures and a multi-role permission system for refined management. Additionally, it provides multi-dimensional statistics on per-user usage, model distribution, and API Key activity, making AI usage traceable and analyzable. The overall design integrates model calls with organizational management, providing foundational support for the large-scale deployment of AI within enterprises.
Telegram Launches No-Code AI Bot Building Tool
Telegram has launched a no-code AI bot building tool, allowing users to obtain tokens via LobsterFather and connect to platforms like Telewer, GPTBots, and Lazy AI to create bots, supporting models such as GPT and Llama.The tool supports configuring conversation styles, task distribution, and multi-bot collaboration, and has been integrated with Telegram Business for use in scenarios such as auto-replies, community management, and customer inquiries. (Techiexpert)
OpenAI Launches Biosecurity Vulnerability Bounty Program for GPT-5.5
According to an official announcement, OpenAI has launched a biotechnology security vulnerability bounty program for GPT-5.5 and is now accepting applications. This program aims to strengthen the safety of its advanced AI capabilities in the biotechnology domain by inviting researchers with experience in AI red-teaming, security, or biosecurity to attempt identifying general jailbreak methods that can bypass its five biotechnology safety challenges.
OpenAI Officially Releases GPT-5.5, Optimized for Agent Tasks and Complex Work Scenarios
According to OpenAI’s official announcement, OpenAI has officially launched GPT-5.5—a next-generation intelligent model designed specifically for handling complex objectives, invoking tools, self-verification, and completing multi-step tasks. The model excels in code writing and debugging, online research, data analysis, document creation, and cross-tool operations. While maintaining response speeds comparable to those of GPT-5.4, GPT-5.5 demonstrates improvements across nearly all evaluation metrics and significantly reduces the number of tokens required to complete equivalent tasks. GPT-5.5 is now available to ChatGPT and Codex Plus, Pro, Business, and Enterprise users. Concurrently, a GPT-5.5 Pro version—optimized for highly demanding tasks—has also been released.