Jon AbbottDevice security has slipped since 2020. Here’s how CISOs can regain control.

Before 2020, most organisations still had a reasonably clear boundary. Corporate data was largely accessed from corporate devices, and the idea of staff connecting their personal laptops to core systems was usually met with a firm no.

The first yes: COVID

Then, almost overnight in March 2020, that boundary collapsed. With offices closed and the business still needing to run, security teams had little choice but to say yes. Personal laptops, home networks and family PCs were suddenly connecting to email, files and core systems.

It was the right call at the time, because the alternative was the business stopping. But many of those emergency decisions were never revisited. What began as a temporary measure became the way we work, extended to contractors and fractional roles, and the controls never caught up.

The second yes: AI

Now security is being asked to say yes again, this time to AI. Every business is under pressure to innovate, and blocking AI outright risks leaving the organisation behind. So new tools, assistants and browser extensions are being adopted at pace, often faster than security can assess them, and frequently on the same unmanaged devices that arrived during the pandemic. The UK’s National Cyber Security Centre backs this up in its latest Annual Review, warning that staff moving company data onto personal devices to use shadow AI “almost certainly” means organisations “lose visibility of how the information is handled and who can access it”.

Why it matters more now

At the same time, the stakes have risen. Our systems run every part of the business, so the pain when they go offline is far greater. That dependency is exactly what makes ransomware so profitable, and criminal and state-sponsored groups have built entire industries around it.

In other words, we loosened control at the very moment the cost of losing it went up. Verizon’s latest Data Breach Investigations Report found ransomware in almost half of the breaches it analysed, a share that’s still rising. And once attackers are in, Mandiant’s M-Trends report shows they typically go unnoticed for two weeks, longer than before. In my experience, unsanctioned software appearing on a device is one of the most reliable signs that a hands-on intrusion is underway.

Saying no to the right thing

Security shouldn’t say no to remote working or to AI. Both are here to stay, and both bring real value. But it is time to start saying no again to something specific: unverified access. Protection requires a level of control. Knowing which sites are being accessed, what applications are installed, and whether endpoint controls are actually working is now table stakes, not a nice-to-have.

This came up at a recent dinner I attended with CISOs. Whenever I raise it, security professionals tend to take one of three stances:

  1. My company simply won’t pay for corporate equipment.
  2. We only allow BYOD for VDI access.
  3. We are fully locked down already.

If you’re in the first camp

I believe this is where most companies sit, and it’s where I want to raise the alarm. Some practical places to start:

  • Make the cost argument. A managed laptop costs a fraction of a single day of downtime, let alone recovery, lost revenue and regulatory fallout. Put those numbers side by side for the board.
  • Set a minimum bar for any device that connects. Check posture before granting access: operating system patched, disk encrypted, security agent running. If it fails, it doesn’t connect.
  • Where you can’t manage the device, contain the data. Managed browsers and virtual desktops can keep corporate data off the personal machine entirely.
  • Be realistic about contractors. You won’t get full management of a third party’s laptop, but verified posture plus contained access is a reasonable and achievable standard.
  • Know what AI is actually in use. Inventory the AI tools, assistants and browser extensions on every device that touches corporate data, sanctioned or not. You can’t set policy for what you can’t see.

If security leaders collectively hold that line, the tide can change.

If you’re in the second or third camp

That’s a great place to be, but “fully locked down” is a claim that only holds on the day you check it. Agents fail silently, exceptions accumulate, configurations drift. And the rush to adopt AI is now testing even the strongest controls, as people install tools and paste data into services nobody approved.

The question isn’t whether you have controls. It’s whether you can prove they’re working today.

Regaining control

This isn’t about undoing what the pandemic changed or slowing down AI adoption. Security can keep saying yes to both. It just needs to know, continuously, that every device touching your data is one it can see and trust.

Jon Abbott is CEO of ThreatAware and a regular speaker at CISO events, including Infosecurity Europe, Black Hat USA and the Cyber Leaders Summit. He’ll be speaking at Black Hat Europe in December.

Secure Every Device in Your Network

Instantly uncover and protect every asset in your IT estate with ThreatAware.

Identify unknown devices, reconcile asset data across platforms, and eliminate security gaps to ensure continuous cyber hygiene.

Book a Demo
App screenshot