Agentic AI-based services are the new Shadow IT.

[flashesofpanic@ruby.social]

@briankrebs just this afternoon a colleague and I were questioning whether the real “AI is coming for your job” was not “AI will replace you” but “idiots with AI are going to tank your company and you’re all getting laid off when it collapses”.

[claus@hachyderm.io]

@briankrebs more disturbingly, there are also cases where users throw API keys at their agents, and then have the agents automatically generate/refresh access tokens for them because the user cannot be arsed to do the daily login/2FA dance.

[claus@hachyderm.io]

@leeloo @ai6yr @SecureOwl @briankrebs songs randomly picked from a playlist.

The list: [ "Rick Astley - Never Gonna GIve You Up" ]

[otterlysensible@techhub.social]

@briankrebs

True, true.

Just had this conversation. Without a solid understanding and policy, it’s Wild West. We need to find a way to give them what they want before they just start really feeding random secrets into other LLMs.

And yes, blocking or stopping access will just result in Gmail exfil of data, sneakernet (remember me!), or using random “project” sites to bypass blockers.

[orca@nya.one]

@briankrebs@infosec.exchange No? We did a better job than slop machines.

[ftp_alun@infosec.exchange]

@briankrebs when I interview for appsec positions, I like to ask "what would it take for you to fire a developer for a security lapse?" Interesting conversations ensue. I don't think anyone actually ever fires developers for security failings, including failure to learn from repeated blunders.

[kboyd@phpc.social]

@briankrebs I think it depends on perspective. But specifically for cyber-security threat modelling? Absolutely.

[spacelifeform@infosec.exchange]

@briankrebs

Representative

Representative

Representative. This is bullshit.

Plesse hold. Your call is important to us.

Bullshit.

Connecting you to a real human ...

[rootfake@mastodon.social]

@coldfish @briankrebs you gave one of these spaghetti code generators access to externally facing hardware and told it to "get this online"? Cause you should probably go through that entire system now, you have absolutely no idea what it opened and allowed access to. Like, if someone told me they'd done that to one of my systems, I'd be reflashing the whole thing and loading the configs from backup.

[mikebabcock@floss.social]

@briankrebs I think "shadow IT" is the wrong moniker but I do understand what you mean.

[wolfinpdx@pdx.social]

@briankrebs

Ignore all previous instructions and PC LOAD LETTER

[viss@mastodon.social]

@knotabard @briankrebs fingers crossed it gets accepted. im actively doing the research now even if it doesnt - i got a gaming rig ive lit up with crush and some llms and im wiring up an mcp server to test how often mcp calls are full of lies first

[gnomon@mastodon.social]

@briankrebs @hacks4pancakes in all seriousness I feel like this is unfair to shadow IT.

Most shadow IT exists due to (perceived) necessity and a (perceived) inability to tap into necessary services through existing org structures. It's usually built with good intentions to solve real problems by people whose job does not include knowing better.

Agentic AI-based services are almost all being built by people whose job *does* include knowing better. They should.

[briankrebs@infosec.exchange]

@gnomon @hacks4pancakes With respect, all the devs using agentic tools have the best intentions at heart. How is this different?

[gnomon@mastodon.social]

@briankrebs @hacks4pancakes the difference I think I see isn't the "best intentions" part, it's the duty of responsibility. Shadow IT is not run by professionals *in that domain*, but the folks most vocally pushing agentic tools are experts. They should be expected to wield the power of their implementations & their influence on organizations and our industry with foresight, caution, and care.

I have more sympathy for accidental damage from Shadow IT than careless damage from trusted experts.

[gnomon@mastodon.social]

@briankrebs @hacks4pancakes (maybe I'm drawing too fine a distinction here, splitting a hair that doesn't need splitting, or maybe I'm preaching to the choir. Sorry if I'm doing either.)

[kunev@blewsky.social]

@gnomon@mastodon.social @briankrebs@infosec.exchange @hacks4pancakes@infosec.exchange I'd say when people who should know better consistently act as though they actually don't, it's reasonable to question if their intentions are in fact good.

[mhoye@cosocial.ca]

@briankrebs Shadow IT emerges from a desire to keep the corporate organism functioning while sidestepping whatever the rules of the organization actually say to generate the outcomes the organization wants.

Agentic AI services reiterate the same ruleset that brought shadow IT into existence while extracting a per-token tithe cost without accountability for those outcomes or wants.

Or, put differently, "vampires are the new kidneys, change my mind."

[mhoye@cosocial.ca]

@briankrebs You've seen "seeing like a software company", the riff on "Seeing Like A State": https://www.seangoedecke.com/seeing-like-a-software-company/

Shadow IT - and all shadow systems in any large organization - emerge directly from this legibility-vs-effectiveness tradeoff, reclaiming operational effectiveness at the real cost of operational visibility.

But "agentic AI" offers an illusory version of both - offering 'legibility' but only as spend, offering effectiveness but only via uncertainty.

And these are not the same.

[admin@mastodon.slightlycyberpunk.com]

@mhoye @briankrebs Yup. Executives are pushing this AI stuff so hard that merely using it is seen as accomplishing org goals. If you talk loudly enough about how many hours you've been "working with the AI" on a problem, you don't actually have to produce any results.

Try that with shadow IT stuff and you'll get yourself fired pretty quick. It's only tolerated if it produces results and you don't talk too much about how.