☆ミ

Developer | Certified Cybersecurity Analyst | Building @GetAevesa | 𓐩 ☀︎ 𝟟𝟟𝟟 𓆩⚚𓆪

[☆ミ @abalous2894]

Traditional security monitors and alerts. That’s fine for humans clicking buttons.

​It’s an absolute disaster when an autonomous agent misreads an MCP tool manifest, fires the wrong API call, or drifts from authorized intent.

​Today, we’re launching Aevesa, the prove layer for the agentic enterprise.

​Our open-core cryptographic engine (Kovera) is still humming under the hood handling delegation chains and receipt primitives.

​But Aevesa is the enterprise platform built around it: hard-stop fail-closed enforcement, compliance mapping (EU AI Act/ISO 42001), and continuous auditability.

​856 files migrated, subdomains mapped, and the control plane is officially live.

​Stop trusting platform-native logs and start recomputing workflow integrity offline.

​Explore the architecture and try our no-login verification demo: https://t.co/17w8D1nq2m

​#AgenticAI #AIGovernance #MCP

[☆ミ @abalous2894]

Finally hot enough in LA to put the AC unit in

[☆ミ @abalous2894]

Finally secured a .com domain with an original, short name with 0 digital footprint from startups and enterprises. A new dawn is on the horizon

[☆ミ @abalous2894]

When you get free credits from @cursor_ai. Goat

[☆ミ @abalous2894]

The enemy of my enemy is my friend. David 🤝 Goliath

[☆ミ @abalous2894]

All the sudden I've got a backup of people that want me to build apps and webpages for them 👀. Where's the AI startup that makes more hours in the day?

[☆ミ @abalous2894]

People are really freaking out about AI's abilities to code and take jobs these past few days. I even saw a post for a software eng job that said it might not be a position next year.

Earlier today I was helping a friend build an app, handed off some design specs to my agent striaght from figma. All the aspects, positions, spacings.

The agent claimed 100% parity was achieved.

The agent was wrong.

Even with the whole design blueprints right before it, the agent still couldn't even get a simple 24px spacing correct. It did 20px and was so proud of itself.

These models are only taking your job if your job is to eat crayons and determine which one tastes best.

[☆ミ @abalous2894]

The gap between an AI vulnerability discovery and full server compromise has collapsed to absolute zero.

In this week's AI Security Incident Report (6/22 - 6/26): The Shrinking Window of Exposure, we track why treating autonomous agents as black boxes is a catastrophic strategy.

The Technical High-Points:
• vLLM Pre-Auth RCE (CVE-2026-22778): A critical 9.8 CVSS flaw chaining malicious video payloads directly to host system takeover.
• HF Transformers Bypass (CVE-2026-4372): Defeating trust_remote_code=False to silently execute untrusted external Python kernels.
• Five Eyes Mandate: Intelligence agencies issue an urgent directive warning that offensive AI has shrunk exploitation timelines from years to months.

Platform logs from Cyera or Databricks are a vendor-scoped autopsy. When Unity AI Gateway denies a tool call, regulated teams need receipts that travel independently.

Stop guessing the compliance path. Prove it.

Read the technical breakdown and test our client-side validation engine completely live (no sign-up required): 👇
https://t.co/UMtytHzPwa

#AISPM #AISecurity #ZeroTrust #AIGovernance

https://t.co/cuohLXVdU1 via @LinkedIn

[☆ミ @abalous2894]

AI-SPM maps what agents can touch.

Auditors ask: when Unity AI Gateway denies a tool call, can you prove it without logging into Databricks, Cyera, or our dashboard?

Platform logs are vendor-scoped autopsy. Regulated teams need receipts that travel.

15 seconds: email with a liability-receipt/v1 JSON → paste at https://t.co/gXGCPLdHVx → green checks. Client-side. No sign-up.

Gateway deny. Cryptographic receipt. Independent verify.

https://t.co/UMtytHzPwa

#AISPM #AISecurity #AI #ZeroTrust

[☆ミ @abalous2894]

@eoinyk LA county itself is more linguistically diverse than Switzerland

[☆ミ @abalous2894]

The spec is frozen and the package compiles standalone. Go run the smoke tests:
​��� GitHub: https://t.co/g3OCtDjsTD
📄 Schema: https://t.co/PN9hruogrZ

[☆ミ @abalous2894]

Palo Alto Networks finalizing their Portkey acquisition proves Big Tech knows AI governance is broken.

​But they’re using a legacy firewall playbook. Mutable proxy logs can be altered. Perimeters don't govern internal execution paths.
​You need cryptographic proof. 👇

[☆ミ @abalous2894]

By separating the open protocol from our proprietary runtime proxy (Genesis), enterprises get ironclad runtime gating with zero vendor lock-in.

[☆ミ @abalous2894]

@sanandreafault LA didn't want to be like NYC and Chicago so it was a law to have buildings less than 13 stories until 1957. Not necessarily the way it is now due solely to earthquakes in the region

[☆ミ @abalous2894]

Starting off your day reporting a bug to a huge tech company. That's my Saturday morning I guess

[☆ミ @abalous2894]

The Wild West era of autonomous AI agents just met its new sheriff. 🛡️

Introducing the Kovera MCP Bridge.

We’re taking agentic security from Trust me, bro; to Cryptographic Attestation.

The Tech Stack:
✅ Zero-Trust Enforcement: Intercepts high-risk tool calls at the protocol level.
✅ Slack-to-Agent Handshake: Managed HITL for high-stakes decisions.
✅ Merkle Ledgering: Every action is anchored to an immutable Sovereignty Receipt.

Don't just watch your agents. Govern them.

#AI #MCP #Security #BuildInPublic #Kovera

[☆ミ @abalous2894]

Observability is for debugging. Kovera is for sovereignty.

By anchoring to $BTC via @opentimestamps , we provide an audit trail that can’t be retroactively altered, even by us.

#AI #CyberSecurity #AgenticAI #Kovera

[☆ミ @abalous2894]

Trust me isn't a security posture for AI agents.

If an agent violates a policy in production, you shouldn't just have a log, you should have a forensic proof.

I’m launching the Kovera Forensic Standard. 🛡️⛓️

[☆ミ @abalous2894]

In this demo, we trigger a high-risk exfiltration attempt in the Vanguard Lab.

Containment: Intercepted mid-call via ZK-architecture.

Governance: Decision pushed to Slack for a human verdict.

Immutability: Every action is anchored to a Merkle-chain and notarized on the Bitcoin blockchain.