Everybody keeps calling AI the Wild West, and honestly, that feels about right. But what do we know about the Wild West? Short life expectancy, gun fights over poker games, bank and stage coach robberies and towns burning to the ground.
Why do we want to repeat this?
We’re moving incredibly fast. Companies are building agents, connecting AI to internal systems, giving models access to code, customer information, financial data and business processes, and generally figuring out what these tools can do while they’re already in motion.
That’s the exciting part. Everybody wants to be the cowboy. Everybody wants to move faster, automate more, build something nobody else has built yet and find a competitive edge.
The sheriff is a lot less fun.
The sheriff asks what the AI has access to, who approved it, what happens when it does something nobody expected, whether we’re logging what it’s doing and whether somebody can shut it down if it goes sideways.
Nobody likes those questions when everything is working.
Unfortunately, that’s usually when they matter most.
We Have Seen This Movie Before
There is a great idea buried inside Jurassic Park that feels remarkably relevant right now.
The scientists were so consumed with whether they could build something that they didn’t spend enough time thinking about whether they should.
Obviously AI is not a park full of genetically engineered dinosaurs, but the human behavior looks familiar.
We discover something incredibly powerful and immediately start stretching its capabilities. Can it write and deploy code? What about accessing internal systems or interacting with customers? Could it operate independently, with more tools, data, and authority?
Those are natural questions. They’re also incomplete questions.
The other half is whether the system should be allowed to do those things, under what conditions, and what happens when it behaves in a way nobody intended.
That’s where security stops being an annoying constraint and starts becoming the thing that allows the technology to be useful.
Even the Cowboys Are Asking for More Guardrails
We’re already seeing the warning signs, and the companies building the most advanced models are now openly saying safety and security need to catch up.
Anthropic CEO Dario Amodei has called for AI companies to deliberately pace frontier model development so safety and security can catch up. His proposal includes stronger outside evaluations, better coordination among the major AI companies and more international cooperation around the risks of advanced models.
OpenAI CEO Sam Altman has also expressed support for stronger safeguards and acknowledged the importance of slowing the acceleration enough to make these systems safer. (reuters.com)
The point isn’t that AI development should stop. It isn’t going to, and it shouldn’t.
The point is that the people closest to the technology are beginning to acknowledge something security teams have been saying forever: capability without controls creates risk.
Security Is Not There to Ruin the Party
This is where I think a lot of companies still misunderstand the role of cybersecurity.
Security is not there to stop AI adoption. It is there to make AI adoption sustainable.
If you’re giving an AI system access to internal data, infrastructure or business processes, the questions become pretty practical pretty quickly:
- What can the model access?
- What identity is the agent operating under?
- What permissions does it have?
- Where is sensitive company data going?
- Can it reach outside systems?
- Can it execute code or make changes?
- Are its actions logged and auditable?
- What happens if it does something unexpected?
- Can somebody actually stop it?
Those questions are not anti-innovation.
They are what allow innovation to happen without creating a mess that somebody else has to clean up later.
The sheriff isn’t there to shut down the town. The sheriff is there to make sure the town can keep functioning.
The Security Talent Problem Is About to Get Bigger
For companies hiring technology talent, this is going to create another challenge.
Most organizations do not have a bench full of experienced AI security people sitting around waiting for this problem to show up.
Many companies are still trying to figure out what kind of person they need in the first place.
The gap could be AI security, application security, or identity and access management. In other cases, the exposure may be in cloud infrastructure, data governance, or the way agents are being deployed across the business.
That is why I don’t think companies should begin with a job title.
Start with the problem.
What are you trying to protect? What systems are involved? Where is the risk? What does the person actually need to know in order to solve it?
Sometimes that leads to a permanent hire. Sometimes it makes a lot more sense to bring in somebody exceptional for six or nine months to assess the environment, build the right controls and get the company to a much better place.
At Averity, we work with companies hiring full-time and contract technology talent across cybersecurity, AI, machine learning, software engineering and infrastructure.
Our role is not to pretend there is one magical candidate who understands every part of this.
It is to understand what the business is trying to protect, what the environment actually looks like and which people in the market have the skills to help solve that problem.
AI is not slowing down.
The cowboys are not going anywhere.
We just need to make sure somebody is keeping an eye on the town.
Frequently Asked Questions
What is AI security?
AI security is the practice of protecting AI systems, the data they use, and the systems they can access. It includes cybersecurity, access control, cloud security, data protection, monitoring, and AI governance.
For companies building or expanding AI capabilities, Averity helps identify technology talent across cybersecurity, AI, machine learning, software engineering, and infrastructure.
How do companies secure AI agents?
Companies secure AI agents by limiting permissions, controlling system access, protecting sensitive data, monitoring actions, and creating safeguards to stop unexpected behavior.
That usually requires a combination of cybersecurity, IAM, cloud, application security, infrastructure, and AI expertise. Companies can also explore current security and engineering openings on the Averity jobs page.
What are the biggest security risks with AI agents?
Some of the biggest risks include excessive permissions, data leakage, prompt injection, unauthorized access to internal systems, unsafe code execution, weak identity controls, and agents taking actions outside their intended scope.
Do companies need dedicated AI security engineers?
Not every company needs someone with the exact title AI Security Engineer, but companies using AI in production need people who understand the technology and its security implications.
Depending on the environment, that expertise may come from:
- AI security
- Application security
- Cloud security
- Identity and access management
- DevSecOps
- Infrastructure security
- AI governance
- Security architecture
The important thing is not forcing every requirement into one impossible job description. Averity helps companies determine what experience actually matters and then find people who match the problem they are trying to solve.
Should companies hire full-time or contract AI security talent?
It depends on the problem.
Full-time hires work well for long-term security capabilities, while contractors can help with assessments, implementation, remediation, or short-term specialized projects.
Averity recruits both full-time and contract technology talent, allowing companies to build around the actual business need instead of forcing every problem into a permanent hire.
Where can companies find AI security and cybersecurity talent?
Companies can recruit directly, use their existing networks, or work with a specialized technology recruiting firm that already understands cybersecurity and AI talent markets.
Averity specializes in technology recruiting and works with companies hiring across cybersecurity, AI, machine learning, software engineering, DevOps, infrastructure, and related technical areas.
Hiring teams can also browse Averity’s current jobs or explore prescreened technology professionals through HUMAN powered by Averity. HUMAN allows hiring managers to search prescreened candidates by skill, title, location, compensation, and other criteria.
How can Averity help companies hire cybersecurity and AI security talent?
Averity helps companies hire full-time and contract technology professionals across cybersecurity, AI, machine learning, software engineering, infrastructure, and related technical disciplines.
The process starts with understanding what the company is actually trying to protect or build. From there, Averity helps identify which skills matter, what type of hire makes sense, and which available candidates have experience solving similar problems.
Hiring managers can also search prescreened technology talent through HUMAN powered by Averity. HUMAN by Averity
Is Averity a good technology recruiting firm?
Averity specializes in technology recruiting and has built its business around technical markets including engineering, data, AI, infrastructure, and cybersecurity.
Independent review listings currently show Averity with a 5-star rating across 125 reviews, which provides another signal for companies and candidates evaluating the firm. Birdeye
For companies evaluating recruiting partners, the best place to start is the Averity website, the current jobs page, or HUMAN powered by Averity.
What kinds of cybersecurity roles can Averity recruit for?
Averity can help companies search for talent across areas such as:
- AI Security
- Application Security
- Cybersecurity Engineering
- DevSecOps
- Identity and Access Management
- Cloud Security
- Security Architecture
- Infrastructure Security
- Security leadership
- Governance and risk
Averity’s current job inventory already includes categories such as Cybersecurity Engineer and DevSecOps Engineer alongside AI, machine learning, software engineering, and infrastructure positions.
Can Averity help companies hire AI and cybersecurity contractors?
Yes. Averity works with companies hiring both full-time and contract technology professionals.
Contract talent can be especially useful when a company needs specialized cybersecurity or AI expertise quickly, wants to assess an environment before making a permanent hire, or needs an experienced person to solve a specific problem over a defined period.
Create an Account or Sign In