Salt Security

Securing the Agentic AI action layer: protecting the APIs, MCP servers, and agents powering modern digital business.

[Salt Security @SaltSecurity]

BIG NEWS! Salt Security’s 1H 2026 State of AI and API Security report just dropped! 🙌 Are you up to date on your AI/API security insights?

Bottom line: You can’t secure AI without securing the APIs that power it.
Download the report: https://t.co/7bELWrKGI8
Read the blog: https://t.co/4KJWxr6Zo7

#cybersecurity #AIsecurity #APIsecurity #AgenticAI #AIagents #SaltLabs #technology

[Salt Security @SaltSecurity]

Time is ticking until the August 2, 2026 EU AI Act enforcement date. Are you prepared?
The EU AI Act is a global business issue. Just like GDPR before it, it reaches beyond EU borders. If your organization does business in the EU, you are in scope.

Watch the full video to learn more: https://t.co/Nnh9q4vBZg

Have questions about AI compliance? Check out the site for more information: https://t.co/d33YRp05JM

[Salt Security @SaltSecurity]

Think you know every AI agent in your environment? Look again.
A static scan finds the agents you know about. Runtime reveals the ones actually operating.

Salt analyzes @awscloud Bedrock invocation logs to uncover unregistered agents, as well as the tools and APIs they use in production.

That’s real agentic visibility. See it in action: https://t.co/8premZM0N2

[Salt Security @SaltSecurity]

AI agents don’t create impact by thinking, they create it by acting.
@huggingface recently disclosed a breach unlike most we’ve seen: the attacker wasn’t a human, it was an autonomous AI agent.

Over a single weekend, the agent exploited code-execution paths, escalated to node-level access, harvested cloud credentials, and moved laterally across internal clusters. More than 17,000 actions were recorded before the activity was flagged.

This wasn’t a traditional attack operating at human speed. It was an agent executing thousands of actions across short-lived environments, using APIs as its primary path to impact.

That’s why protecting the model alone isn’t enough. Organizations need visibility and behavioral baselines across the Action Layer to identify when agents access unauthorized systems, misuse credentials, or move laterally at machine speed.

Salt is built for that visibility gap: mapping the complete path from models and tools to APIs and downstream systems, identifying dangerous connections, and detecting when runtime behavior deviates from its intended purpose.

The Hugging Face incident wasn’t an anomaly, it was a warning.
Read the latest from Salt 🧂 → https://t.co/315vTgPiFJ

[Salt Security @SaltSecurity]

Black Hat is almost here, and we’re bringing some of the brightest minds in AI security with us 🌐

Be sure to book a meeting with the Salt team while you're in Las Vegas. You'll have the opportunity to connect with two of Salt's leading security researchers and get firsthand insights into the latest AI & API threats, emerging attack trends, and what organizations can do to stay ahead → https://t.co/RXB9JEoyBl

Whether you're looking to strengthen your security strategy, discuss real world research, or see how Salt helps secure the AI & APIs that power modern business, we'd love to connect.

[Salt Security @SaltSecurity]

Your AI agents shouldn’t be invisible to security.
Salt now automatically discovers agents deployed in Amazon Bedrock, including their tools, prompts, and configurations.
No manual inventory. No waiting for traffic. Just continuous visibility into the agentic attack surface.
#AgenticSecurity #AmazonBedrock @awscloud

[Salt Security @SaltSecurity]

Salt has officially landed in Australia 🇦🇺
With locally hosted infrastructure in AWS Sydney and a dedicated in-country sales + technical team, we’re bringing Agentic and API Security closer to the enterprises moving fastest on AI.

Local data. Local expertise. Global-scale protection.
This is how Salt moves: quickly, strategically, and wherever innovation goes next.

Learn more: https://t.co/RkarPluaJM

[Salt Security @SaltSecurity]

Have questions about the EU AI Act? Whether you're just getting started or preparing for enforcement, our experts will answer your questions live on July 28th.

Sign up for our upcoming webinar: https://t.co/lyrkX4Tt5J

[Salt Security @SaltSecurity]

We’ve reached the big 100! 🎉
With 100 pre-built policies, the Salt Policy Hub gives security teams an immediately deployable governance layer across APIs, MCP servers, agent permissions, compliance, and runtime behavior.

AI moves fast. Security governance should move faster.

See how Salt is building security for the entire agentic ecosystem: https://t.co/pZ35FtYbcK

[Salt Security @SaltSecurity]

Something big is rolling into Las Vegas 👀🚚
Keep your eyes on the streets during Black Hat... you just might spot Salt Security lighting up the city in a way you can't miss!

See what we have going on: https://t.co/M8zfUYYPlG
Not going to Black Hat? Catch Salt at one of our roadshows: https://t.co/W9KL8TIpW3

[Salt Security @SaltSecurity]

11 emerging MCP security risks published today, and every one of them is an API security problem.
MCP connections = API calls. No visibility into agent API traffic = blind spot attackers are exploiting.

Read @securityblvd's piece here: https://t.co/Z0vjT4NrvD

[Salt Security @SaltSecurity]

Meet Salt V2: a faster, smarter security ecosystem built to keep teams moving.
Find anything in two clicks. See how agents, MCPs, capabilities, and APIs connect. Understand risk instantly. Turn posture gaps into ready-to-use fixes.

This is what progress looks like 😎
and we’re just getting started.

Read all about it here: https://t.co/pM39q5sCno

[Salt Security @SaltSecurity]

Are you wondering how to secure your org's evolving Agentic threat landscape? Well, we have the event for you.

Come join Salt Security’s exclusive Agentic Security Workshop at Shangri-La, Shard, London, on the 6th of August 🌐

Seats are limited so book your spot today!
Register here: https://t.co/71fxHWwBR3

[Salt Security @SaltSecurity]

Attackers are scanning for your MCP servers and AI agent credentials right now.
49+ IPs, structured recon.
MCP = access to your internal APIs, DBs, file systems.

Most orgs have zero visibility into this traffic, but Salt catches it🧂

Read up here @SANSInstitute: https://t.co/6o6kVPGghv

[Salt Security @SaltSecurity]

Researchers weaponized AI hallucinations into a botnet: pre-register the fake package names LLMs invent, wait for agents to pull & execute. 85-100% success rate.

How do you prevent this? Salt Security flags the anomalous API calls agents shouldn't be making 🛑

Read more @SecurityWeek: https://t.co/6A58aJrkY4

[Salt Security @SaltSecurity]

Attention all #cybersecurity enthusiasts! Salt Security is #hiring 😎
We’re expanding our Customer Success team and searching for dynamic Technical Account Managers to join us.

Please share with your network or apply here: https://t.co/TEBmxpbdBA

[Salt Security @SaltSecurity]

In the agentic era, you can’t secure what you can’t see, making Salt Cloud Connect for AWS an absolute game changer.
Organizations can instantly discover APIs, MCP servers, and agentic activity across @awscloud environments (without deploying sensors or routing traffic).

Check how it’s done → https://t.co/QYhEP5HbRO

[Salt Security @SaltSecurity]

Come say hey in Mandalay Bay!
Salt Security will be at Black Hat and we’d love to chat! 😎
Swing by Booth #5938 to:
- Get a live demo
- See your Agentic Graph
- Talk through how Salt is redefining Agentic Security
- Learn about Salt Code

We have super fun activities you don’t want to miss! Check them out here: https://t.co/M8zfUYZnbe

[Salt Security @SaltSecurity]

@getaigovernance @nudge_security @oasissec @silverfort @SnapLogic @CycodeHQ @Hacker0x01 @hiddenlayersec Thank you for sharing! 🎉

[GetAIGovernance @getaigovernance]

Best AI Security Platforms 2026 Expert Guide

mindguard @nudge_security @oasissec @SaltSecurity @silverfort @SnapLogic WitnessAI @CycodeHQ @Hacker0x01 @hiddenlayersec

https://t.co/CN2qWK7ARG

[Salt Security @SaltSecurity]

@ForbesTechCncl Thank you for sharing! This is a must-read💡