Salt Safety used the stage at AWS re:Invent this week to unveil two main enhancements to its API Safety Platform, introducing a generative AI interface powered by Amazon Bedrock and increasing its behavioural menace safety to safeguard Mannequin Context Protocol (MCP) servers by way of AWS WAF. The bulletins spotlight the corporate’s rising give attention to visibility, threat discount and real-time defence in more and more advanced cloud and AI environments.
On 1 December, Salt launched “Ask Pepper AI”, a pure language interface designed to assist safety groups immediately question their whole API property. Constructed on Amazon Bedrock, the instrument permits customers to ask plain-English questions (resembling “Which of my APIs expose PII?” or “What APIs have the very best Threat Rating?”) and obtain rapid, actionable insights drawn from Salt’s API Discovery, Posture Governance and Menace Safety capabilities.
With organisations struggling for readability in sprawling cloud environments, Salt’s H2 2025 State of API Safety Report discovered that solely 19% really feel “very assured” within the accuracy of their API stock, whereas 15% admit they have no idea which APIs expose private information. Salt says “Ask Pepper AI” helps shut these gaps by democratising entry to essential safety data and accelerating each incident response and threat prioritisation.
“API safety is advanced, however understanding your threat shouldn’t be,” mentioned Michael Nicosia, Co-Founder and COO at Salt Safety. “‘Ask Pepper AI’ makes it easy. By utilizing Amazon Bedrock, we’re placing highly effective, intuitive safety insights into the fingers of everybody from SOC analysts to CISOs. When most organisations aren’t even positive what their API stock seems to be like, the flexibility to only ask and get a right away reply is a game-changer.”
Two days later, Salt introduced a second main functionality: the extension of its patented API behavioural menace safety to detect and block malicious intent concentrating on MCP servers. MCP servers enable LLMs and autonomous brokers to execute duties by calling APIs and instruments, however their rising utilization has outpaced safety controls. Usually deployed with out central oversight and uncovered to the web, they’re changing into a brand new goal for attackers searching for entry to delicate information and system performance.
Constructing on Salt’s just lately launched MCP Finder know-how, the corporate now permits organisations to establish misuse or abuse of MCP servers and routinely block threats utilizing AWS WAF, leveraging real-time behavioural intelligence from the Salt platform.
“Most organisations don’t even know what number of MCP servers they’ve, not to mention which of them are uncovered or being abused,” mentioned Nick Rago, VP of Product Technique at Salt Safety. “This functionality lets them take motion shortly, utilizing current controls to stop actual threats without having to deploy new infrastructure.”
By combining MCP discovery with AWS WAF enforcement, clients can block assaults earlier than they affect purposes, uncover shadow or unmanaged MCP situations, prolong edge safety to the AI motion layer, and constantly replace defences as attacker ways change.
