Ibrahim Sukari

Founder, Entrepreneur and Pokemon Collector

[Ibrahim Sukari @cyberibrahim]

@forloopcodes What a joke

[Ibrahim Sukari @cyberibrahim]

Lapsus$ just hit Vodafone.

VODAFONE TELECOMMUNICATIONS

Major telco breach underway. #CyberSecurity #Lapsus #Vodafone

[Ibrahim Sukari @cyberibrahim]

Still, with non-technical teams now generating and shipping prod code (even with reviews), and a flatter org with fewer engineers overall, the review burden seems heavier on a leaner team.

In crypto, where one subtle vuln can lead to exchange drain, deep security expertise + ruthless scrutiny has always been non-negotiable.

Hoping the guardrails and processes truly raise the security bar.

[Ibrahim Sukari @cyberibrahim]

@nandantwts Protect them from being stolen

[Ibrahim Sukari @cyberibrahim]

@rez0__ I agree. Not responsible disclosure. This is like giving the keys while the vault is open.

[Ibrahim Sukari @cyberibrahim]

This week AI cooked cybersecurity:

Claude Security → Cursor Security → GPT-5.5 Cyber.

All in 7 days.

The robots are now auditing the robots.

What a time to be alive

[Ibrahim Sukari @cyberibrahim]

GitHub Copilot switching to usage-based AI Credits… and yeah, good luck if you’re doing a lot of agentic stuff.

Meanwhile, OpenAI is still keeping ChatGPT Plus and Pro on straightforward monthly subscriptions with pretty generous limits. No surprise bills for most people. Smart move holding the line while everything else is getting more expensive.

Who else is thinking about canceling Copilot?

[Ibrahim Sukari @cyberibrahim]

Classic over-privileged AI agent + brittle infra.

1. CLI token gave full delete rights (no scoping)
2. Destructive op via one command, no gates
3. Backups in same deletable volume
4. Prompts/Plan Mode failed as expected

LLMs + broad creds = self-inflicted insider threat.

Audit tokens, add hard gates, separate backups. Prompt safety isn't security.

(Agents aren't "rogue" - your controls were.)

[JER @lifeofjer]

https://t.co/ofucbVgkLV

[Ibrahim Sukari @cyberibrahim]

@IntCyberDigest Amateur hour

[Ibrahim Sukari @cyberibrahim]

This is actually a smart take on AI security. Handing real credentials to unpredictable agents feels like playing with fire (hello, prompt injection nightmares). Their proxy setup keeps secrets hidden, brokers the calls, adds proper scoping + rules + full audits.

Way better least-privilege approach.

Open source.

Security folks, worth a test run.

[Infisical @infisical]

Any secret an agent can read is a secret an attacker can steal.

So we built the fix: Agent Vault, an HTTP credential proxy and vault for AI agents.

Secret managers were built for deterministic services. They return credentials to the caller and trust them to behave.

AI agents break that assumption. They are non-deterministic, prompt-injectable, and increasingly sitting in front of your prod APIs and databases.

Instead of returning credentials directly to the agent, Agent Vault forces the agent to proxy requests through it, brokering credentials at the proxy layer and forwarding requests to any target API all in an interface-agnostic way. Credentials stay in the vault, encrypted with AES-256-GCM. The agent never touches them.

What you get:
→ Brokered access through HTTPS_PROXY, not retrieval. Nothing to exfiltrate.
→ Firewall-like access rules implemented at the proxy.
→ Multi-vault RBAC to scope agents to a tight blast radius.
→ Full audit trail and inspection of every passing call.
→ All compacted in a single Go binary executable; available as a Docker container.

Read out announcement post for a further breakdown: https://t.co/8UP3BoCbin

Try it → https://t.co/i8DDJ1dtou

[Ibrahim Sukari @cyberibrahim]

Morning at #BlackHatAsia2026 in Singapore.

Violet Blue’s keynote hit hard: “Privacy is the Captain. Security is the Practice.”

She showed how privacy failures are often the real starting point for big breaches, and why data sovereignty efforts in APAC offer a smarter path than extractive models.

Thought-provoking way to kick off Day 2 👏

#BHASIA #DataPrivacy #Cybersecurity

[Ibrahim Sukari @cyberibrahim]

@elonmusk This is what proper in-flight internet looks like.

Australian carriers, stop making us suffer with laggy WiFi and get Starlink already!

[Ibrahim Sukari @cyberibrahim]

@budapp Let's go

[Ibrahim Sukari @cyberibrahim]

@SpaceX @cursor_ai This is huge.. Pairing Cursor with Colossus could supercharge secure code generation and vulnerability detection for everyone building critical systems.

Cyber defenders will move faster than ever. Net positive for the industry, excited to see what comes next.

[Intrix Cyber Security @CyberIntrix]

Cyber threats are evolving fast and so should your defence.

If you’re unsure how exposed your business might be or where to start, that’s usually the best place to begin the conversation.

Explore how we can support your business: https://t.co/keYvcT93G2

[Ibrahim Sukari @cyberibrahim]

@DiffeKey Too little too late unfortunately. The damage has been done.

[Ibrahim Sukari @cyberibrahim]

Poor security meets pro-level gaslighting.

Accidentally re-exposed sensitive chats (with creds and all), researchers flag it via HackerOne… and they close it as “intended behavior.” Then the first statement: “No breach, just unclear docs.” Classic.

Next time, maybe private-by-default before the drama?

[Marius Avram @securityshell]

Holy shit… the exploitation of CVE-2025-55182 has reached a new level. There’s now a publicly available Chrome extension on GitHub that automatically scans for and exploits vulnerable sites as you browse. Absolutely wild. 🤦‍♂️

[Ibrahim Sukari @cyberibrahim]

Our platform has put a quick checker together to see if your org domain has been affected by Shai Hulud 2.0
https://t.co/aqy0nxm6l9

[Ibrahim Sukari @cyberibrahim]

Our team is analyzing the repos from the #ShaiHulud NPM fiasco. We are seeing email addresses from LARGE organisations like:

* Accenture
* DocuSign
* Cloudflare
* Ericsson
* Nutanix
* Phillips
* SAP
* Thomson Reuters
* Vodafone
* https://t.co/jEmW7bLKEM

Page coming up soon to check your email and org.