mmissoldtw ←
Nudge Security485 posts
Nudge Security avatar

Nudge Security

@nudge_security

Nudge Security helps modern organizations achieve effective SaaS security and governance at scale.

◷ Joined February 2022▧ 182 Photos and videos
Nudge Security avatar
Nudge Security@nudge_security·14 Jan 2025

📣 Today's press release recaps highlights from 2024, which include more than doubling our customer base and tripling ARR! Thank you to our customers, investors, partners, and friends who have provided invaluable feedback and support along the way. 🙏🏻 bit.ly/40umBu6

Post attachment
Nudge Security avatar
Nudge Security@nudge_security·Nov 12

An AI agent doesn’t have one identity. It borrows one per connection, mostly OAuth grants from a person, and any of them can resolve to a personal account. Endpoint discovery tells you what’s running. Not as whom. bit.ly/4zIbhtD

Nudge Security avatar
Nudge Security@nudge_security·Nov 11

Block a risky tool and plenty of people just find a way around it. Now you can’t see the very risk you meant to manage. Calibrated friction: observe, inform, justify, escalate. Save the block for the risks that earn it. bit.ly/3UqE6eL

Nudge Security avatar
Nudge Security@nudge_security·Nov 5

"Allowed in my org” and “right for my org” aren’t the same answer, but a lot of AI agents stop at the first one. An agent that approves access without knowing who’s actually requesting it isn’t saving you work. It’s just moving the risk somewhere else. bit.ly/4sF2kOK

Nudge Security avatar
Nudge Security@nudge_security·Oct 29

A CMDB tells you every VM you run. It tells you nothing about who can read your customer data. Our 2027 prediction: the attack surface isn’t a list of assets anymore. It’s a graph of grants, created in a few clicks with no sign-off. bit.ly/4i8o2as

Nudge Security avatar
Nudge Security@nudge_security·Oct 27

A vendor gets approved for one narrow use case. 18 months later it’s got broad access, company-wide adoption, and AI features nobody reviewed. The paperwork still says low risk. bit.ly/4sjhQjn

Nudge Security avatar
Nudge Security@nudge_security·Oct 16

Only 30 to 40% of SaaS and AI apps ever get a formal TPRM review. The ones that do are assessed once, then left alone. Adaptive risk management adjusts controls as risk changes, not when the next review comes due. bit.ly/3VJBjxo

Nudge Security avatar
Nudge Security@nudge_security·Oct 14

OAuth grant numbers worth sitting with: 88 per employee, 31 with data-level access, ~45 min to review each manually. Do that math at scale. Manual review was never going to keep pace. bit.ly/4zwm21y

Nudge Security avatar
Nudge Security@nudge_security·Oct 12

Gartner: half of all SaaS breaches will trace to overprivileged OAuth tokens by 2027. You can’t review a grant you don’t know exists. Dormant grants and stale integrations don’t show up in logs. Wrote about it for @BleepinComputer bit.ly/4i6OYat

Nudge Security avatar
Nudge Security@nudge_security·Oct 8

Manual review of one high-risk OAuth grant: 45 minutes. Verdict: revoke. Same grant through our new AI agent: 15 seconds. Same verdict. Average employee has 88 OAuth grants. 31 touch data. Nobody is reviewing those by hand. This will help ➡️ bit.ly/4yCwVPu

Nudge Security avatar
Nudge Security@nudge_security·Oct 7

TPRM didn't break because of AI. It was already broken. We assessed vendors, never usage. We screened on scores that measured their marketing site, not your data. We triaged and hoped. The score is about them. Your exposure is about what you gave them. bit.ly/3U7KKGr

Nudge Security avatar
Nudge Security@nudge_security·Oct 6

Most vendor reviews are accurate for exactly one day. After that: new OAuth grants, teams, AI features, & agents. Nothing triggers a second look. Adaptive Risk Management in Nudge Security re-scores risk as it changes & tells you what to fix first. bit.ly/4rOpNNg

Nudge Security avatar
Nudge Security@nudge_security·Oct 5

Three questions before connecting any AI tool: what can it see, what can it do, who owns it? Good framing. The catch: those questions get answered on a consent screen, by one person, right before they click Allow. Not in a review. bit.ly/4dVq9Mb