Everyone is getting hacked. Are you next?

Answer 10 questions. Get a one-page report on where AI leaves you exposed, and your first three moves.

Start the free check

It’s free, takes about 2 minutes and needs no email address. Your answers stay in this browser.

Book a call

AI security readiness report

Larkmere Pay (fictional sample)

You may be more exposed than you think.

Leaks

Exposed

Agents

Exposed

Break-ins

Partly defendedLower confidence

Biggest risk from your own AI

An AI agent could do damage you can’t trace or undo.

Biggest risk from attackers’ AI

AI makes phishing cheap and convincing, and it can capture one-time codes as well as passwords.

Your first three moves

  1. Log every agent action and make the risky ones reversible.
  2. Move staff sign-ins to passkeys or security keys.
  3. Give each agent its own identity and the least access it needs.

Self-reported estimate, not an assessment

It’s already happening.

  • Leaks · April 2026

    Vercel suffered a security breach that began with the compromise of a third-party AI tool used by one of its employees, and some customer configuration data stored by Vercel was compromised.

    Vercel security bulletin
  • Agents · July 2026

    When tested, OpenAI’s models were able to break out of the isolation and gain access to Hugging Face’s systems, and OpenAI says that it all happened without any human direction.

    OpenAI incident report
  • Break-ins · November 2026

    The FBI has warned of cyber attackers leveraging an open-source AI tool to run intrusions, with some taking less than 3 hours between targeting and gaining privileged access.

    FBI FLASH notice
Riccardo Spagni

Riccardo Spagni, CEO of Kala

Riccardo leads every Sprint himself, from the first interview to the final briefing.

  • Lead maintainer of Monero from 2014 to 2019
  • Builder of Blink, a browser-use agent
  • Named inventor on 17 granted US patents
  • Former CTO of Tari Labs, the organisation behind Tari and Yat
More about Riccardo

Kala AI Security Readiness Sprint

Is your board asking whether AI has made you a target? The Sprint is a 10-day AI security assessment, personally led and independent of model and cloud vendors. It ends with a prioritised 90-day plan to close your gaps.

Kala AI Security Readiness is a Kala service. Kala also sells payment, policy, and trust infrastructure products. Kala AI Security Readiness engagements never recommend them.

Starting at US$25,000

The final fee depends on scope (the teams, systems, and agents involved) and is fixed in writing before work starts.

Choose what to do next

Start the free check

It’s free, takes about 2 minutes and needs no email address. Your answers stay in this browser.

Book a call

AI Security Readiness Check

This free AI security check keeps your answers in this browser. Click or tap an answer to move on.

0 of 10 answered

Select a terrace to see its questions. Each one rises as you answer.

01 / 10   Your viewHow well is your company defended against AI security threats?
02 / 10   LeaksDo you know which AI tools your staff use, and with what data?
03 / 10   LeaksWhat can your AI vendors do with the data you send them?
04 / 10   LeaksDo you know which outside apps and AI integrations can reach your systems?
05 / 10   AgentsWhat can AI agents, including coding assistants, do in your systems without approval?
06 / 10   AgentsIf an AI agent made a costly mistake tomorrow, could you trace and undo it?
07 / 10   AgentsCan your AI read email or web pages and then take actions?
08 / 10   Break-insCould a cloned voice or fake video call get a payment approved?
09 / 10   Break-insIf an attacker stole a staff password today, what would stop them?
10 / 10   Break-insHow fast would you spot and contain a break-in?
How the scoring works
  • Average the answered questions in each pillar. Below 1.5: Exposed. From 1.5 to below 2.5: Partly defended. At least 2.5: Well defended.
  • Not sure is missing data, not zero. One unknown lowers confidence. Two or more show Not enough answers.
  • The overall position averages the pillar averages that have a level. Q1 is not scored. The headline compares it with the overall position.
  • The biggest risk from your own AI is the lowest Leaks or Agents answer. The biggest risk from attackers’ AI is the lowest Break-ins answer.
  • Moves follow those two risks, then the lowest remaining answer. Unknown questions fill empty places.
  • Ties follow A2, A1, L1, B2, B1, A3, L3, B3, L2.
  • A score of 3 uses the no-gaps line.
  • Levels describe good security practice, not a test of your systems.
All ten questionsQ1. How well is your company defended against AI security threats?L1. Do you know which AI tools your staff use, and with what data?L2. What can your AI vendors do with the data you send them?L3. Do you know which outside apps and AI integrations can reach your systems?A1. What can AI agents, including coding assistants, do in your systems without approval?A2. If an AI agent made a costly mistake tomorrow, could you trace and undo it?A3. Can your AI read email or web pages and then take actions?B1. Could a cloned voice or fake video call get a payment approved?B2. If an attacker stole a staff password today, what would stop them?B3. How fast would you spot and contain a break-in?

Which one do you need? The free check gives a self-reported estimate in minutes. The Sprint is Riccardo’s paid assessment, and it ends with a 90-day plan.

AI security readiness report

Larkmere Pay (fictional sample) · 2026-11-20

Self-reported estimate, not an assessment

You may be more exposed than you think.

Where you stand

Leaks

Exposed

Agents

Exposed

Break-ins

Partly defendedLower confidence

Biggest risk from your own AI

An AI agent could do damage you can’t trace or undo.

Biggest risk from attackers’ AI

AI makes phishing cheap and convincing, and it can capture one-time codes as well as passwords.

Your first three moves

  1. Log every agent action and make the risky ones reversible.
  2. Move staff sign-ins to passkeys or security keys.
  3. Give each agent its own identity and the least access it needs.

What this can’t tell you

Self-reported answers are unverified. Levels describe good security practice, not a test of your systems.

You answered Not sure to:

How the scoring works

Talk through your report with Riccardo and decide whether the Sprint fits.

Book a callFree, 30 minutes and confidential.

The Sprint is Riccardo’s paid assessment. It takes 10 business days and ends with his own pillar ratings and a 90-day plan.

Starting at US$25,000

The final fee depends on scope (the teams, systems, and agents involved) and is fixed in writing before work starts.

The report is gone once you close this page.

Send the check to a colleagueChange answers

Kala AI Security Readiness Sprint

Find your AI security gaps before attackers do.

When your board asks if AI has made you a target, a customer asks how you secure your AI, or an agent is about to get real access, you need answers that hold up.

The Sprint is an AI security assessment. It's a deep dive into where your data goes, your AI agent permissions and your resilience to AI attackers and it's led personally by Riccardo Spagni, CEO of Kala. In 10 business days you'll have a prioritised plan for the next 90 days that your board can back.

Starting at US$25,000

The final fee depends on scope (the teams, systems, and agents involved) and is fixed in writing before work starts.

Book a call

Free, 30 minutes and confidential.

Confidential, independent and personally led

Riccardo will lead the Sprint himself. This ensures that the assessment will be done in an independent way, not depending on a specific model or cloud vendor. The Sprint will be executed under a mutual NDA with Kala. All findings will only be disclosed to the persons as indicated by you.

Kala AI Security Readiness is a Kala service. Kala also sells payment, policy, and trust infrastructure products. Kala AI Security Readiness engagements never recommend them.

Three pillars

Where is your data going?

Leaks

What can your AI do without asking?

Agents

Could you stop an attacker who uses AI?

Break-ins

How the scoring works

Deliverables

  • Executive findings memo
  • Pillar ratings: Riccardo’s own rating of the company on Leaks, Agents, and Break-ins
  • AI tool, data-flow, and connected-app inventory
  • AI and agent risk register
  • Agent authority map
  • AI-assisted attack scenarios
  • Prioritised 90-day plan
  • Final executive briefing

How it runs

  1. Map where AI touches your data, systems, and money.
  2. Trace what a mistake or an attacker could reach.
  3. Rank the gaps by damage and cost to close.
  4. Deliver the 90-day plan and brief the executives or the board.
Everything the Sprint covers
  • Executive interviews
  • AI usage and shadow-AI inventory
  • Data-flow and AI vendor terms review
  • Connected-app, integration, and access-key review
  • Agent identity, authorisation, approval, audit, and rollback review
  • Prompt-injection exposure review
  • AI-amplified threat analysis
  • Payment and impersonation control review
  • Sign-in and credential review
  • Detection and response walkthrough
  • Priority scoring
  • 90-day execution roadmap
  • Executive or board briefing

Before you book

Who is this Sprint useful for?

The Sprint is designed for leadership teams of funded technology companies, as well as companies whose success depends on technology, and which possess data, money, or systems that a malicious actor may want to obtain. The Sprint is most useful for organisations that have an executive sponsor who can act on the Sprint’s findings. The Sprint is not a penetration test, a compliance audit, a security monitoring service, or a means of hiring security staff.

Is this a penetration test?

No. The Sprint is a review of your actual situation. Riccardo will not attack you but will work from interviews, documents and settings of your own systems. He will not certify you against a certain standard but will advise you, if necessary, what you need to do with a penetration test.

How much time from my team will this take?

The Sprint primarily consists of interviews with leadership and system reviews with the people that run your systems. The specific people that will be needed will be outlined in the written scope of work.

What access will Riccardo need to my company?

He’ll need access to the leaders in your company, as well as access to documentation and the people that run your systems, so that he can review your data, system architecture, permissions and agents. This work will be done remotely and under a mutual NDA prior to commencing.

What happens after the 10th day?

You’ll receive all of the deliverables listed above, as well as a final briefing to your executives or board. After that, if you need help executing on the plan, Riccardo is happy to assist with the next phase of work.

What if this isn’t a good fit?

Schedule a free call to discuss your needs, and if it isn’t a good fit, Riccardo will let you know on the call before agreeing to any fees.

After the Sprint

Riccardo also offers follow-on work: carrying out the plan, interim leadership, deeper security reviews and board briefings. Ask about them on the call.

Book a call

Free, 30 minutes and confidential.

Riccardo Spagni

CEO of Kala

Riccardo Spagni

The Sprint is personally led by Riccardo Spagni, CEO of Kala, from the first interviews to the executive briefing. Riccardo has built, defended and sold software businesses. His work spans cryptography, distributed systems and AI agents.

Book a call

Free, 30 minutes and confidential.

Experience

Book a call

Book a free, confidential 30 minutes call to discuss your organisation’s AI-related security exposure, and how a Sprint could help. Please bring your report with you if you have one!

  1. Fill out the form and send it
  2. Expect a message from aisecurity@kala.to, and choose a time that fits you
  3. Talk for 30 minutes with Riccardo to understand if this is a good fit, and what it would entail
Engagement confidentiality

Every engagement starts only after both sides sign Kala’s mutual NDA. Findings go only to the people you name.

Privacy notice

Don’t include secrets, credentials, source code, non-public security vulnerabilities, legal admissions or regulated personal data.

Kala Holdings Inc. uses these details only to arrange the call and discuss fit and scope. Cloudflare and Google process them for Kala, and they are never shared for advertising.

Enquiry retention

Kala deletes enquiries 12 months after the last contact. If an engagement follows, its terms apply instead.

aisecurity@kala.to

Request sent

A reply from aisecurity@kala.to will set a time for a free, 30-minute call to determine fit and scope.

Engagement confidentiality

Everything you share on this form and during the call will be kept confidential. If you choose to engage Kala, no work will commence until you’ve signed Kala’s mutual nondisclosure agreement, and Kala will only share its findings with you and other people you designate.

Start the free check

It’s free, takes about 2 minutes and needs no email address. Your answers stay in this browser.

Back to home

Something went wrong

Your request may not have been sent. Your details are still in the form. Try again, or email aisecurity@kala.to.

aisecurity@kala.to

Privacy notice

Last updated 20 November 2026

Kala Holdings Inc., a company incorporated in Delaware, USA, is the data controller of this site. You can contact them on aisecurity@kala.to for anything to do with privacy.

When you submit the “Book a call” form, Kala receives your name, work email address, company, role, the topic you are interested in, when you would like the call (if you specify it) and any note you make. Kala uses this data to respond to your enquiry, set up a call with you and discuss if there is a fit and what the scope of that fit might be. Kala treats this data in confidence and will never use it for advertising.

Kala’s justification for collecting this data is “Taking steps at the data subject’s request prior to entering into a contract”, and “Legitimate interests”, specifically dealing with business enquiries and ensuring the form is not abused.

Cloudflare is used to host this site, run the Turnstile security check on the form and email enquiries to Kala. Your IP address is used as part of this security check. Google hosts the inbox that enquiries are sent to. No database of enquiries is held by Kala and the site does not log what you enter into the form. Both Cloudflare and Google may process your data outside of your country, including in the USA.

The site does not use analytics, ad related trackers or record your session.

Your enquiry will be deleted 12 months after your last contact with Kala. If you decide to engage with Kala then the terms of that engagement will apply.

The data that you enter into the AI Security Readiness Check, including your answers, the results and the report, never leaves your browser. It is never sent to or stored by Kala. The PDF file is created on your device and as soon as you close the page the report is deleted.

You can request a copy of the data that Kala holds on you, ask for it to be corrected or deleted by emailing aisecurity@kala.to. You can object to how your data is used. If you are not satisfied with the response you receive then you can take your complaint to your local data protection regulator, such as the Information Regulator in South Africa, the Information Commissioner’s Office in the UK or your local authority in the EU.

Back