API Security: Fix the Endpoints Attackers Find First
API security is the work of finding every live endpoint, controlling who can call it, and watching its traffic for abuse. Salt Security reported in 2025 that 99% of surveyed organizations had API security issues in the prior year, while Thales recorded more than 40,000 API incidents in the first half of 2025. Inventory, authentication, and monitoring close that gap.
By Abdullah Shafique, AI Engineer
An application programming interface, or API, is the code that lets one application ask another for data or an action. A mobile app signs a user in through an API, a checkout page charges a card through one, and a partner system pulls order status through one. When any of those connections is weak, the API hands over the data or performs the action, because that is what it was built to do.
What is API security, and why does it matter now?
API security covers the full life of an endpoint, from the day a developer ships it to the day it is retired: discovery, authentication, authorization, rate limits, input checks, logging, and monitoring. The reason it matters now is scale. In Salt Security's Q1 2025 report, nearly one-third of organizations reported API growth of 51 to 100% over the prior year, and one-quarter reported growth above 100%. Salt Security, Q1 2025 The H2 2025 report put it in portfolio terms: 42% of organizations managed between 101 and 500 APIs, and 14% managed more than 1,000. Salt Security, H2 2025
Why do attackers go after APIs?
Attackers go after APIs because the traffic looks normal. Thales, reporting on its Imperva customer telemetry, recorded more than 40,000 API incidents in the first half of 2025 across more than 4,000 monitored environments, an average of over 220 per day. APIs made up only 14% of the attack surface in that data, yet they drew 44% of advanced bot traffic. Thales API Threat Report, H1 2025
Salt Security's Q1 2025 traffic analysis found that 95% of API attacks over the prior 12 months came from authenticated sources, and 98% targeted external-facing APIs. A stolen token, a partner account with too much access, or a logged-in user probing another customer's records all pass the login check, and the attack hides inside traffic a firewall expects to see.
Tim Chang, Vice President Application Security Products at Thales, described that shift in plain terms: "APIs are the digital economy's connective tissue – but that also makes them its most attractive attack surface." Thales API Threat Report, H1 2025 His point was that attackers now bend business logic instead of relying on malware, so the traffic looks legitimate.
The Thales data also shows what that abuse looks like in practice. Data scraping accounted for 31% of API bot activity, and coupon and payment fraud accounted for 26% of attacks. Credential stuffing and account takeover attempts rose 40% on APIs without adaptive multi-factor authentication. Remote code execution probes made up 13% of attacks, with Log4j, Oracle WebLogic, and Joomla among the most targeted flaws.
Which API flaws cause the most damage?
Salt Labs examined attack techniques in customer environments for its Q1 2025 report and mapped 80% of attack attempts to the OWASP API Security Top 10 list. Misconfiguration came first by a wide margin, and broken object-level authorization came second.
| Flaw | Share of observed attacks, Salt Q1 2025 | What it means in practice |
|---|---|---|
| Security misconfiguration | 54% | Default settings, verbose errors, or missing security headers left uncorrected. |
| Broken object-level authorization | 27% | A caller changes an object ID, such as an order number, and reaches another customer's record. |
| Broken user authentication | 1% | Login checks fail, a small share in this dataset. |
| Security monitoring and logging failures | 1% | Traffic is not logged, or no one watches the logs. |
Asked about production API challenges, respondents named vulnerabilities at 37%, sensitive data exposure at 34%, and API authentication weaknesses at 29%. Salt Security, Q1 2025 The largest losses come from endpoints that were misconfigured, over-permissioned, or left unmonitored.
How do you find shadow and zombie APIs?
A shadow API is an endpoint that exists and runs outside the official, monitored channels of an organization. A zombie API is an older version that should have been retired and was not. A team cannot protect an endpoint it does not know exists, and an attacker scanning for forgotten endpoints will find it first. Thales reported that organizations typically have 10 to 20% more active APIs than they are aware of. Thales API Threat Report, H1 2025
The Salt reports show how common that blind spot is. In Q1 2025, only 15% of respondents expressed strong confidence in their API inventories, 34% lacked visibility into sensitive data exposure through APIs, and only 20% continuously monitored APIs. By H2 2025, only 19% were very confident in their inventory accuracy, and 54% relied on developer documentation to identify sensitive data exposure. Salt Security, H2 2025
A working inventory comes from live evidence. Gateway logs show which endpoints receive traffic, code repositories show which routes are defined, and cloud accounts show which services are exposed. Together they reveal the gaps: endpoints with traffic but no owner, versions with traffic but no support, and internal endpoints reachable from the internet. Retire what is unused, assign an owner to what remains, and record what data each endpoint can return.
What should an API security program include?
Salt Security defines API posture governance as consistent security standards and frameworks across the whole API ecosystem, so gaps get fixed before attackers use them. In Q1 2025, only 10% of organizations had such a strategy in place, while 43% planned to implement one within the next 12 months. Salt Security, Q1 2025 The H2 2025 report urged the same shift: continuous discovery, stronger governance, runtime protection, and safeguards for AI-generated code.
In practice, a program has five parts. Discovery keeps the inventory current. Authentication and authorization decide who can call each endpoint and which records they can touch, with object-level checks on every request that names a record. Rate limits and bot controls slow credential stuffing and scraping. Runtime protection inspects live traffic for abuse of business logic. Monitoring ties it together, with logs that someone reviews and alerts that fire while an attack is still running.
Two related StackIntel guides cover the detection and cloud sides of this work: extended detection and response, which correlates signals across endpoints, identities, cloud, networks, and email, and cloud security posture management, which covers misconfigured cloud settings.
How do you start this week?
Start with the inventory, because every later control depends on it. Pull live endpoints from the gateway, compare them with the documented list, and investigate every endpoint in traffic but not in documentation. Retire test endpoints and old versions that still answer requests, and note the owner, authentication method, and data returned for each endpoint that stays.
Next, check authorization on the sensitive endpoints first: account records, payment data, and admin actions. Test whether one authenticated user can reach another user's records by changing an ID in a request. Where that test succeeds, add an authorization check on the server for every request. Add rate limits to login and checkout endpoints, require adaptive multi-factor authentication where credential stuffing is a risk, and turn on logging for every external endpoint. Set a review date for the inventory itself, because an inventory checked once will be wrong again within a quarter.
Nearly every surveyed organization had already hit API security issues, and the incident count in the first half of 2025 alone passed 40,000 in one vendor's telemetry. The teams that fare better treat APIs as part of the attack surface from the start: discovered, owned, limited, and watched.
Frequently asked questions
How many APIs does a typical company manage?
In Salt Security's Q1 2025 report, 43% of organizations managed up to 100 APIs and 34% managed between 101 and 500. The H2 2025 report found 42% managing between 101 and 500 APIs and 14% managing more than 1,000.
What is a shadow API?
A shadow API is an endpoint that exists outside an organization's official, monitored channels. Thales reported that organizations typically have 10 to 20% more active APIs than they are aware of.
Why does authentication alone not stop API attacks?
Salt Security's Q1 2025 traffic analysis found that 95% of API attacks over the prior 12 months came from authenticated sources. A valid login does not prove a caller should reach a particular record, so authorization checks still have to run on every request.
What is API posture governance?
It is a consistent set of security standards applied across the whole API ecosystem. In Q1 2025, only 10% of organizations had such a strategy in place, while 43% planned to implement one within the next 12 months.
How many API incidents were recorded in the first half of 2025?
Thales recorded more than 40,000 API incidents in the first half of 2025 across more than 4,000 monitored environments, averaging over 220 per day.
Comments
Post a Comment