The top data breaches of 2025 so far

Paradox: attackers are becoming both more visible and harder to spot. Ransom notes and leaked sites often make headlines, while quiet mistakes and stolen tokens put sensitive data at risk for years.

This list of the biggest data breaches in 2025 keeps things straightforward, focusing on the facts. We’ll continue updating it until the end of the year as new numbers and investigations come in.

Key facts from 2025’s breaches

1. Many recent data breaches started with identity abuse or third parties, which turned one weak control into a massive incident.
2. Quiet exports and misconfigurations moved stolen data at scale without setting off alarms.
3. Healthcare stayed the costliest sector, with long-lived records and operational pressure that cannot pause.
4. Ransomware attacks increasingly copied data first, then encrypted systems to add pressure.
5. Legacy servers, broad permissions, and stolen credentials gave bad actors simple paths in.
6. SaaS tokens and analytics beacons exposed personal information even without a network intrusion.
7. Basic data security steps, such as phishing-resistant multi-factor authentication (MFA) and strict vendor isolation, would have prevented or shrunk many events.

The top 12 data breaches of 2025

These are the biggest security failures of the year so far. Each one shows how simple mistakes or overlooked gaps can expose millions of people.

1. PowerSchool data breach 2025: student records and Social Security numbers

How it happened: In January 2025, someone claimed they had copied data for more than 62 million students and nearly 10 million teachers from PowerSchool’s systems. A few months later, Texas confirmed that 880,000 state residents were caught up in the breach. Investigators say the attackers got in using a contractor’s stolen login, which opened the door to school records nationwide.

By September, the Texas Attorney General filed a lawsuit against PowerSchool, calling it a major data breach that put minors’ personal information at risk. In some states, public filings mention names, contact details, and Social Security numbers. In other areas, school transportation and medical information may also have been exposed.

Why it matters: PowerSchool isn’t just one app; it’s a student information system that schools use to track enrollment, grades, bus routes, and family contacts. When attackers get inside, the damage spreads fast. Districts that link PowerSchool to other apps may now have to rotate passwords, check logs for strange activity, and clear out old accounts before those links become the next problem.

2. SK Telecom SIM authentication data breach: stolen data on a national scale

How it happened: In April 2025, South Korea’s biggest telecom provider admitted that attackers had stolen authentication records tied to USIM (Universal Subscriber Identity Module) cards. A USIM is the tiny chip inside a phone that lets the carrier verify your identity on the network. Regulators later said malware had been present in SK Telecom’s systems for years before the company caught it.

The exposed records could allow criminals to clone SIM cards, hijack text messages, and take over online accounts. Officials described the case as nationwide in scale, with 26.96 million customers affected. The government fined the company and ordered it to improve controls and monitoring.

Why it matters: SIM data may not sound dramatic, but it’s powerful. Banks, email providers, and social apps often send log-in codes by SMS. If attackers can duplicate a SIM, they can grab those codes and bypass account protections. Even partial leaks become dangerous when combined with sensitive data already traded on the dark web.

3. Salesforce CRM data breach: stolen credentials and OAuth tokens

How it happened: In September 2025, the FBI issued a warning about two groups it tracks, UNC6040 and UNC6395. Both were quietly stealing customer data from Salesforce tenants. One group used vishing—phone calls that trick employees into giving up multi-factor codes—and then ran Salesforce’s own Data Loader tool, which is normally used by admins to bulk import and export data.

The other group relied on stolen OAuth tokens, which are digital keys that let connected apps (like Salesloft or Drift) act on a user’s behalf without asking for another password. With those tokens, the attackers could query Salesforce data directly. Multiple companies later confirmed they were affected and notified customers.

Why it matters: This wasn’t ransomware with loud lock screens. It was the quiet theft of CRM records: names, phone numbers, emails, business contacts, and support tickets. Because many companies use the same third-party apps, one compromised token can ripple across dozens of organizations. What looks like a small slip can easily become another entry in the growing list of recent data breaches.

4. U.S. Treasury vendor tool intrusion: bad actors and third-party vulnerability

How it happened: In late December 2024, the U.S. Treasury learned that cybercriminals had broken into a contractor’s remote support tool. These tools let outside vendors log in and fix systems as if they were sitting inside the office. In this case, the attackers used that door to reach unclassified Treasury computers and documents.

The incident became public around the New Year and carried over into 2025, with investigations pointing to a group linked to China. It matters for a data breach 2025 list because this vendor tool is everywhere. A single weak login or too much access can give outsiders control over a whole agency.

Why it matters: One small help desk session can turn into control of hundreds of machines. Businesses that rely on contractors face the same risk. Stronger logins, verified devices, and limits on what vendor tools can touch are key to reducing exposure.

5. Oracle Health legacy servers compromised: healthcare vulnerabilities

How it happened: At the end of March 2025, hospitals reported that threat actors had broken into old Cerner “data migration” servers. These were outdated, on-premises systems that some facilities still used to move patient information into newer platforms. Oracle Health later confirmed that these obsolete servers were compromised, but insisted its cloud systems were not.

Notices through the summer named at least 14,485 affected individuals, though the real number is likely higher. Because Cerner is connected to many hospital environments, the breach spread risk far beyond one location. Even when only old servers are hit, the mix of names, IDs, and medical details creates a long-term privacy headache.

Why it matters: Health data lasts for life, and attackers know it. Old servers and forgotten accounts often sit unpatched, waiting for someone to exploit them. One weak system at a big vendor can ripple through dozens of hospitals at once.

6. Yale New Haven Health data breach: exposed personal information

How it happened: On April 11, 2025, Yale New Haven Health said hackers had been inside its systems since March 8. The disclosure put 5.56 million people on the federal breach portal. The leaked files included demographic details and other protected health information, the kind of data that fuels scams and identity theft.

Patients quickly filed lawsuits, and by late summer, the health system was already in settlement talks. For context, 5.5 million people are equivalent to the population of a small U.S. state. Losing that much personal information in one go makes this one of the biggest hospital breaches of the year.

Why it matters: Hospitals can’t simply turn off systems to fight cybercriminals. They must continue treating patients while responding, which makes containment harder. That constant pressure leaves cracks open, and attackers exploit them.

7. Blue Shield of California analytics leak: sensitive data in ad systems

How it happened: On April 9, 2025, Blue Shield of California admitted that a misconfigured Google Analytics setup had been quietly leaking member data to Google Ads for years. No cybercriminals were involved. Instead, a tracking tool meant for marketing accidentally sent private information to advertising systems.

About 4.7 million people were affected. While full medical charts and payment cards weren’t exposed, browsing histories, plan details, and claims data slipped out. That information can still be valuable for profiling and targeted ads.

Why it matters: Not all breaches involve a break-in. Sometimes the problem is a tool left in the wrong place with the wrong settings. These hidden leaks rarely set off alarms, which makes them even harder for companies to spot.

8. DaVita ransomware attacks with stolen records

How it happened: In mid-April 2025, DaVita, a large dialysis provider, reported a major ransomware incident. The attackers not only encrypted systems but also exfiltrated patient records. Federal data later showed that about 2.7 million individuals were affected.

Reports indicated laboratory databases among the compromised systems. Clinics kept operating, but the exposure of medical details and identifiers created lasting privacy and fraud risks.

Why it matters: Modern ransomware is a two-step threat: copy the data first, then lock the systems. Even when backups restore operations, the stolen files often resurface on leak sites or the dark web. For healthcare providers, the cost is not just downtime but years of monitoring and dealing with lawsuits.

9. Qantas customer data exposed by vendor: data security lessons

How it happened: Between July 2 and July 9, 2025, Qantas confirmed that a third-party contact center platform exposed data on 5.7 to 6 million customers. The information included names, emails, phone numbers, birth dates, and loyalty numbers. The airline emphasized that no passport or payment details were part of the breach.

The fallout was serious enough that executives saw bonuses cut, a measure aimed at showing accountability. For attackers, travel data is highly useful. It allows phishing campaigns that mimic real flights or loyalty offers, luring customers into giving away even more information.

Why it matters: Airlines rely on many outside providers for customer support, booking, and loyalty programs. A single weak vendor can compromise passengers across the network. This case mirrors other recent data breaches, where the company in the headlines wasn’t the one operating the vulnerable system.

10. TeleMessage government communications exposed: archives and cyber threats

How it happened: On May 4, 2025, TeleMessage, a vendor that archives government communications, was breached. Collections of its data later appeared on activist sites that host leaked material. Reports said the exposed trove included plaintext message fragments, metadata, and government contact details, such as Customs and Border Protection addresses.

TeleMessage exists to meet legal record-keeping rules by storing copies of secure communications. But when those copies are kept in a less secure form, they become easier targets than the original encrypted apps. One point of entry can open an entire archive.

Why it matters: Archives centralize sensitive material. Without encryption, strict access, and short retention, archives become easier targets and raise cyber threats for every connected agency.

11. United Natural Foods systems disrupted: supply chain data breach

How it happened: On June 5, 2025, United Natural Foods filed an 8-K with the SEC saying it had shut down parts of its IT systems after a cyber incident. The disruption slowed order processing and food distribution, raising concerns about revenue and operations. For a supply chain that runs on tight margins and schedules, even short outages ripple quickly to store shelves.

The company did not release numbers on stolen or exposed records. Still, the case belongs in a data breach 2025 roundup because it shows how cyber incidents in logistics affect the public even when the scope of stolen data isn’t fully clear.

Why it matters: Logistics firms run on thin margins and tight schedules. Attacks here often combine data breach and downtime, which raises real-world impact even when record counts are not yet known.

12. Marks & Spencer ransomware after social engineering

How it happened: In April 2025, Marks & Spencer faced a major disruption when threat actors used social engineering to gain access and launch ransomware. The attack crippled click-and-collect services and disrupted operations for weeks. By September, reports put the financial impact at around £300 million in lost profit.

Attribution bounced between different groups, but the tactics were familiar: convincing staff to hand over access that attackers then used to deploy ransomware and steal data. The retailer did not disclose how many records were affected, but the scale of the business disruption was clear.

Why it matters: Retail depends on always-on systems, from store checkouts to online orders. Once attackers seize accounts, they can shut down fulfillment and extract customer data in the same move. Preventing session hijacking, enforcing phishing-resistant MFA, and limiting employee permissions can block the kinds of tricks used here.

What these incidents say about current cyber threats

1. Identity is the soft spot, and third parties widen the blast radius

• A stolen contractor login unlocked PowerSchool’s nationwide student records.
• OAuth tokens and vishing drained Salesforce CRMs quietly.
• Qantas and the U.S. Treasury felt the cost of vendor tools that function like master keys.

About one-third of breaches involve a partner or supplier. When access moves through help desk platforms, call centers, or SaaS connectors, every permission you grant becomes shared vulnerability.

2. Data theft has gone quiet, at scale

• Salesforce campaigns focused on extraction first.
• DaVita showed copy first and encrypt later.
• TeleMessage proved that archives, kept for compliance, can expose leaked data when stored carelessly.
• Blue Shield of California showed how tracking tags moved member data for years.

Ransom demands cluster in the low six figures. As more victims refuse to pay, attackers increasingly resort to data-only extortion and multi-party pressure tactics.

3. Healthcare remains the costliest and one of the most exposed sectors

The breaches at Yale New Haven Health, Oracle Health legacy servers, and DaVita show why: healthcare organizations have sprawling ecosystems, care that cannot pause, and long-lived records that keep high value to attackers. Average breach costs in healthcare sit far above the global average, and vendor issues keep reappearing.

4. Tech debt fuels initial access and creates vulnerabilities

Legacy servers, dormant accounts, weak MFA, and stolen browser cookies gave attackers simple paths. Infostealers and session hijacking turned browsers into skeleton keys, since many devices mix personal and work accounts on unmanaged endpoints. Edge gear with late patches added more vulnerabilities.

The picture is clear: bad actors pursue the path of least resistance—stolen identities, permissive tokens, neglected systems, and misconfigured analytics.

Defensive moves that actually help

Executives do not need slogans. They need to make the blast radius smaller and remove clear vulnerabilities. The patterns across these 2025 data breaches point to direct controls that reduce cyber threats without stalling teams that ship products or care for patients.

1. Use phishing-resistant MFA (FIDO2/WebAuthn). Enforce number matching where keys can’t reach. Disable legacy/basic authentication, require modern OAuth, and add Conditional Access tied to device health and location. Add SIM-swap and port-out PINs. Re-auth on sensitive actions for high-risk roles.
2. Shrink the third-party blast radius. Inventory every vendor tool and remote-support path. Grant least privilege, expire access by default, and isolate vendors on monitored jump hosts. Put incident SLAs and logging in contracts. Use ZTNA and segment so vendors can only reach the apps they service.
3. Control SaaS exports and tokens. Audit connected apps, revoke stale OAuth tokens, shorten lifetimes, and narrow scopes. Rate-limit or approval-gate bulk exports; alert on abnormal CSV/API pulls. Restrict Data-Loader-type tools to break-glass admins and monitor with DLP.
4. Retire or isolate legacy systems. Remove internet exposure, require MFA for admin access, patch on schedule, and block management interfaces from the open internet. Scan weekly for forgotten portals and outdated firmware.
5. Harden browsers and sessions. Require Device Posture Security checks; block access from non-compliant devices. Shorten session lifetimes, prefer device-bound tokens, and use a Secure Web Gateway with DNS Filtering by category, Web Protection, and Download Protection.
6. Stop quiet exfiltration and ransomware. Deny outbound to new destinations, watch object-store writes, and cap egress per identity. Seed honeytokens. Protect backups with MFA, immutability, and network separation; test restores. Disable internet macros, restrict remote admin tools, and enforce allowlists.
7. Remove SMS from recovery. Use email or hardware-key recovery, add telecom account locks for execs and finance, alert on SIM changes, and re-validate sessions on number changes.
8. Fix analytics leaks. Ban PII in analytics, govern tags with an allowlist, block beacons on protected paths, run synthetic tests, prefer server-side tagging with redaction, and enforce a strict CSP.
9. Secure archives. Encrypt with HSM-backed keys, restrict decryption by role, shorten retention, isolate connectors, disable plaintext exports, and log every query and export.
10. Segment always-on operations. Pre-stage network segments, separate admin IT from clinical/operational systems, keep read-only EHR failover, train for downtime, and alert on abnormal bulk record queries.

How NordLayer helps protect against data breaches

The biggest 2025 data breaches show repeating weak points: stolen credentials, over-permissive vendor tools, and misconfigured systems that exposed sensitive data. To counter these, organizations need controls that reduce the chances of leaked data and shrink the impact if attackers get in.

NordLayer offers several tools that directly address these risks.

• Phishing-resistant MFA and single sign-on (SSO) integrations help block stolen credentials from being reused, a common trigger in recent data breaches.
• Cloud LAN creates secure device-to-device connections across your organization, enabling remote collaboration while isolating traffic. This helps limit the potential impact if one device or vendor account is compromised.
• DNS Filtering by category and Web Protection prevent employees from reaching malicious sites that deliver infostealers and ransomware, cutting off one of the main sources of stolen data that fueled the 2025 breaches.
• NordLayer’s Business VPN, powered by AES-256 and ChaCha20 encryption, provides secure network access for employees. It ensures that sensitive data stays protected in transit, even if attackers intercept traffic.

Together, these measures reduce the opportunities attackers exploited in many of the 2025 breaches. They reduce the chance that bad actors can exploit simple vulnerabilities, plant ransomware, or move stolen data to the dark web.