Can AI Apps Really Be Secure From Birth? Oh Yes (it's an Atsign of the times)
Timing and IT are not always aligned – spoiler alert: major understatement…
How many times over the decades have we seen a wonderful technology looking for a problem to solve? Some are still out there, still looking for that elusive problem – and they were 8-bit applications 😊. But that’s certainly true on far more occasions than we encounter our “holy grail” experience of a right here, right now (I think there’s a song in there somewhere) technology that actually DOES solve a very real and contemporary problem.
Yes, I’m talking about AI. And DevOps. I have lots of developer mates; they love creating, they hate debugging and finishing their work of code art, (you know who you are) in order for it to be actually “enterprise-ready”. Documenting? You’re having a larf… But worst of all – “security” is scarcely in their digital vocabulary. Meantime, we have company boards reading daily about how AI is now the ultimate business maker or breaker and pressuring IT to embrace and convert to AI dev, terrified they will otherwise lose their market share. Or disappear completely. Apart, that is, from the CISO who thinks: “releasing a raft of untested AI apps with zero security into our infrastructure and supply chain mechanisms – really?”
Yes, it’s the ultimate stand-off; where our good old Welsh mate “Dai Chotomy” looms large. In some scenarios, it’s just FUD-factor noise-making, designed to force boards into spending on technology that they don’t actually need (that would be the technology looking for a problem scenario). However, In this case, the need to create and release secure AI apps as rapidly as possible is very real. And – trust me on this one – Sooty and his magic wand doesn’t do the trick (sorry, US readers – Google it). We used to talk about the need for DX or Digital Transformation (were these marketing guys dyslexic?). Well, that window of opportunity has long been boarded up, along with those businesses – especially retail – that failed to transform. Now we have AI – the real DX agent, even if the initialisation really doesn’t work this time.
So… how do we resolve this security vs AI demand conundrum? Well, we do have some basic acknowledgement of security playing its part in AI dev not least, of course, Project Glasswing. However, this Mythos security blanket has only just started to be sown together. Meantime, Anthropic itself has announced Claude Code Security, currently available in preview form and designed to scan codebases for security vulnerabilities and suggest targeted software patches for human review. It sounds remarkably like a basic detection scanner for AI code, which is hardly proactive and is still using the “reduce the attack surface” mentality as its provenance. Solution? No.
How about, then, a real solution that starts with a zero-attack surface (no TCP ports or IP addresses to attack) and secures any AI app natively, regardless of the LLM being used? Dream on, you say but, no – it does exist. And I’ve tested it. It’s from Atsign, who I first encountered via the Tech Trailblazers awards (wot I am a judge on) and it won – of course. I say, of course because the technology is bloody brilliant. It’s the classic “why didn’t someone think of it before?” scenario. Well, now it’s finished, released – Atsign Platform – along with a simple, no-code, dev tool that we proved takes weeks of dev time down to one morning: Atsign AI Architect – and is ready to revolutionise AI development and release cycles – overnight, or even over breakfast (not the all-day variety). We’ve already walked through several, very different use cases for the technology but the only actual limit is your imagination – old cliché but true in this case. Moreover, regardless of the nature of the app, the process, from conception to release, is basically the same – every time. You can even use multiple LLMs to create, for example, different front ends for different device platforms, for the same app.
Suffice to say, being simultaneously ground-breaking and disruptive (in a very good way) I had no option but to give it a gold award. So, rather than simply recycling the report contents here, just head straight to the report itself. It’s a free download and you don’t have to have your kids and parents taken hostage in order to get hold of a copy – just click on the link:
https://exclusive.atsign.com/broadband-testing-report/ – and prepare to be impressed. And then start writing a wish list of all the applications you’d like to transform into AI versions, confident in the knowledge that they will be totally secure on release – and that release will be very rapid. And then just imagine and create a whole load more.
Have your eCake and eat it? Oh yes…