Microsoft has unveiled MDASH, a multi-modal AI security platform that outperformed Claude Mythos Preview on the CyberGym benchmark. According to Microsoft's blog post, MDASH scored 88.4% compared to Mythos Preview's 83.1% in vulnerability discovery capabilities. The platform, also known as Microsoft Security multi-modal agentic scanning harness, represents the first multi-modal service to be included in the CyberGym benchmark, an AI security benchmark developed by UC Berkeley's Center for Responsible, Decentralized Intelligence. MDASH consists of over 100 specialized agents that hunt for vulnerabilities and debate whether flaws discovered are real or exploitable, demonstrating Microsoft's approach to agentic system design rather than relying on a single model. The platform has already surfaced 16 previously unknown vulnerabilities in its first run against Windows operating system, including four critical remote-takeover flaws addressed during this month's Patch Tuesday.
At Consensus 2026 in Miami Beach, Binance Chief Security Officer Jimmy Su warned that attackers are increasingly using AI to speed up exploits and reduce the time between vulnerability discovery and attack launch. As reported by BeInCrypto, Su explained that what used to require a team of five or six red teamers to find vulnerability can now be done with one person from my AI tool over a weekend. The technology has fundamentally changed the attack landscape, enabling faster scaling while creating new vulnerabilities that defenders must counter with advanced AI tools. According to BeInCrypto reports, Jimmy Su, Chief Security Officer at Binance, discussed how artificial intelligence is fundamentally changing the landscape of cryptocurrency security, noting that over the last six months, AI adoption has been expanding across the entire business, but especially in security.
Microsoft's MDASH represents a multi-agent, multi-harness system that leverages model diversity to avoid lock-in to particular AI models. As reported by Microsoft, the platform uses 100 different sub-agents that can find vulnerabilities and provide necessary patches, with customers able to pick and choose which models they want to enable from available options including GPT 5.5, 5.6, 5.5-Cyber, Sonnet and Opus. Taesoo Kim, VP of Security Research at Microsoft, explained that Microsoft is pretty uniquely positioned because of this model diversity that we are leveraging as part of MDASH, enabling the company to pick and choose the best available model for various security tasks. This approach contrasts with single-model solutions like Claude Mythos, which has dominated cybersecurity conversations since Anthropic's Project Glasswing announcement. The adoption of MDASH is rapid, with everyone internally at Microsoft leveraging MDASH for their security tests, including the Windows team that has integrated the tool for the entire build process and pipeline.
Google Threat Intelligence Group (GTIG) documented the first confirmed case of a threat actor deploying an AI-generated zero-day exploit in an active operation — a Python script targeting a two-factor-authentication bypass vulnerability in an open-source web administration tool. The exploit's characteristics (extensive docstrings, structured formatting, educational inline comments) were distinctive of LLM-generated code. PRC-nexus actor UNC2814 was documented using "persona-driven jailbreaking" — elaborate researcher personas to extract vulnerability research from commercial AI models — as a standard operational workflow. This disclosure represents the operational threshold-crossing from AI as an experimental adversary capability to AI as an integrated component of nation-state attack chains. The combination of UNC2814's documented Google Sheets C2 channel (used to spy on 53 telecoms and governments in 42 countries) with the persona-driven jailbreaking workflow gives this PRC actor a full chain: AI-extracted vulnerability research → AI-assisted exploit generation → established C2 infrastructure for deployment at scale.
As reported by BeInCrypto, Su emphasized that even though crypto prices are trending higher over time, the real story is in increasing adoption itself. He highlighted growing rural use of cryptocurrency, tokenization of different rural assets, and the convergence of traditional finance and cryptocurrency products. According to BeInCrypto, Binance has recently added significant new features, including contracts to trade oil prices, signaling the company's movement toward traditional finance features. At Consensus 2026, Su announced that Binance is developing tokenized stocks and commodity tokens, with these tradfi-style offerings intended to attract more traders and expand the exchange's addressable market, while increasing competition with conventional brokers. This expansion into tradfi products is expected to increase Binance's competitor pool while potentially capturing a larger market share.
As reported by BeInCrypto, Su outlined how AI tools are being used defensively to synthesize signals from different areas including email networks, endpoints, and device logs. The technology helps security teams analyze logs more broadly and in-depth, enabling faster threat identification. Binance is also using AI for cloud coding, which can synthesize entire attack chains from vulnerability discovery to deployment. According to BeInCrypto reports, Su emphasized that Binance continues to balance user experience with protection through AI tools, with the company developing context-aware security features that recognize trusted devices and user behavior patterns. Binance plans to introduce more AI-powered features in the coming months, including enhanced user authentication and risk-based challenge systems, as the company navigates the evolving threat landscape where AI has lowered the barrier for attackers to find and exploit vulnerabilities.
At Consensus 2026, Su pointed to recent U.S. Senate activity on the Clarity Act as part of a broader trend toward clearer rules that are encouraging banks and traditional financial firms to offer crypto services. He said adoption, not short-term price swings, is the defining trend for the industry over the next five to ten years. The regulatory environment is creating opportunities for institutional participation, with Binance's product expansions and security upgrades arriving as institutional interest in crypto grows and regulators consider frameworks to bring banks and funds into the market. This regulatory clarity is expected to significantly expand Binance's addressable market and competitive landscape as traditional financial institutions enter the cryptocurrency space.
