I broke AppLovin's mediation cipher protocol

TL;DR

A researcher has decrypted AppLovin’s mediation cipher, revealing that the encrypted payload contains enough device data to re-identify iPhones across different apps without user consent. This challenges assumptions about privacy protections in ad tech.

A researcher has successfully decrypted AppLovin’s mediation cipher protocol, revealing that the encrypted bid requests include sufficient device data to deterministically re-identify iPhones across different applications, even when users deny App Tracking Transparency (ATT). This development raises significant privacy concerns and questions about the effectiveness of current user tracking restrictions.

The researcher captured and decrypted thousands of encrypted ad bid requests from AppLovin, discovering that the payloads contain detailed device information. The encryption scheme relies on a static salt embedded in the SDK and a publisher-specific SDK key, combined with a predictable timestamp-based counter, resulting in ciphertext that can be decrypted with the right key. The decrypted data includes a comprehensive set of device attributes, such as hardware details, OS version, and unique identifiers, which can be used to re-identify devices across multiple apps regardless of ATT status.

The encryption scheme employs a non-authenticated cipher based on a variation of SplitMix64, which does not provide cryptographic security against tampering. The timestamp used in encryption is embedded in the ciphertext, allowing the researcher to recover the exact time of each request, further aiding device fingerprinting. The decrypted payloads include both device fingerprint data and opaque tokens sent to ad networks, with some signals readable and others encrypted, but all containing identifiable device information.

Why It Matters

This discovery challenges the assumption that ATT is the only barrier to user fingerprinting. The decrypted data shows that ad networks can re-identify users across apps without relying solely on IDFA, undermining privacy protections and potentially violating user expectations and regulations. It also exposes vulnerabilities in AppLovin’s encryption approach, which lacks cryptographic integrity features like MAC or AEAD, making it susceptible to tampering and spoofing.

Amazon

device fingerprinting privacy protection

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Background

Prior to this, the industry widely believed that ATT was the primary means for user identification and tracking. AppLovin’s SDK is one of the largest mediation platforms, involved in billions of ad requests daily. The encryption scheme used by AppLovin was assumed to protect user data, but the researcher’s findings reveal that the encryption is weak and that device fingerprinting remains feasible through the encrypted payloads. Similar concerns about device fingerprinting and privacy have been raised before, but this is the first known breach of AppLovin’s specific cipher protocol at this scale.

“The encrypted bid request carries enough device data to deterministically re-identify the same iPhone across apps from different publishers, even when user denies ATT.”

— Researcher

“The cipher used by AppLovin is not cryptographically secure, and the inclusion of timestamps in the ciphertext makes device fingerprinting straightforward.”

— Privacy expert

Amazon

iPhone privacy shield case

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

What Remains Unclear

While the researcher has decrypted thousands of envelopes and demonstrated the feasibility of device re-identification, it remains unclear how widespread or systematically exploited this vulnerability is within the industry. AppLovin has not yet publicly responded to these findings, and it is unknown whether they will update their encryption protocols or implement additional security measures.

Amazon

encrypted device ID blocker

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

What’s Next

Next steps include further analysis to determine if other ad networks use similar encryption schemes, industry response from AppLovin and regulators, and potential updates to SDK security practices. Researchers and privacy advocates will likely scrutinize the encryption weaknesses and push for stronger cryptographic protections in ad tech.

Amazon

privacy screen protector for iPhone

As an affiliate, we earn on qualifying purchases.

As an affiliate, we earn on qualifying purchases.

Key Questions

Does this mean my device can be tracked across apps even if I deny ATT?

Yes, according to the researcher’s findings, device fingerprinting via decrypted payloads can re-identify iPhones across apps without relying on IDFA or ATT.

What specific data can be used to re-identify devices?

The decrypted data includes detailed device info such as hardware model, OS version, screen dimensions, and unique identifiers, which can be combined to fingerprint a device.

Has AppLovin responded to these findings?

As of now, there has been no public statement from AppLovin addressing the decryption or its implications.

Will this affect other ad networks or just AppLovin?

The findings specifically relate to AppLovin’s encryption scheme, but similar vulnerabilities could exist in other ad tech SDKs if they use comparable encryption methods.

You May Also Like

AI Trading Bots: Promise or Peril for Retail Investors?

With AI trading bots promising speed and efficiency, are retail investors risking more than they realize—find out what you need to know.

Khosla Ventures is betting $10M on Ian Crosby, whose last startup, Bench, imploded

Khosla Ventures leads a $10 million seed round for Ian Crosby’s new startup Synthetic, aiming to develop fully autonomous AI bookkeeping, despite past startup challenges.

AI Governance Tokens: Letting Algorithms Vote?

Spot how AI governance tokens empower algorithms and stakeholders but also raise questions about transparency, fairness, and the future of AI oversight.

AI‑Generated NFT Collections: Creativity Meets Code

AI-generated NFT collections blend cutting-edge algorithms with blockchain technology to create unique,…