Is Tor still safe to use?
Claims that law enforcement has de-anonymized at least one Tor hidden service user have reignited debate over how safe Tor really is and for whom. Commenters note that the reported attack targeted an outdated client lacking newer protections, and that most successful deanonymizations still hinge on endpoint exploits, traffic analysis, or user opsec errors rather than a wholesale break of Tor’s design. Many conclude Tor remains one of the strongest practical tools for online anonymity—but only within a clearly defined threat model and with realistic expectations that state-level adversaries may still succeed against high-value targets.
Context: German Investigation & Ricochet Case
- Discussion centers on a German TV report alleging a timing-based deanonymization of an onion service (Boystown CSAM forum) between ~2019–2021.
- Tor Project blog says: likely a guard-discovery attack on a user of an old Ricochet client lacking Vanguards(-lite); mitigations have existed in Ricochet-Refresh since 2022.
- Technical details from NDR/CCC are limited or not shared with Tor; several commenters flag this lack of transparency as a problem.
“Is Tor Safe?” – Threat Model Framing
- Many stress “safe for whom, against whom?”
- Against local ISPs, adtech, basic law enforcement, and many non-Western regimes: Tor is seen as significantly safer than plain internet or most VPNs.
- Against well-resourced Western intelligence (NSA, Five Eyes, Mossad, etc.): several argue you should assume they can sometimes deanonymize users, especially targeted ones.
Guard Discovery, Timing & Correlation Attacks
- Known weaknesses: traffic analysis using packet size/timing, guard discovery, and flow-correlation, especially for onion services repeatedly contacted.
- Vanguards (entry + middle guard design) aim to make these attacks more expensive but not impossible; mitigations shift economics, don’t “fix” the problem.
- Some propose additional obfuscation (VPNs with padding/shaping, more hops, aggressive padding), but others note performance and practicality issues.
Relay/Exit Node & Global Adversary Concerns
- Long debate about whether governments run “most exits” or a large fraction of relays; concrete evidence is scarce, but metrics show many relays in 14‑Eyes countries.
- Tor community does active bad-relay detection and relay governance; still, commenters consider large-scale relay control or IXP fiber taps realistic for state actors.
- Important nuance: exits are irrelevant for onion services; their main risk is correlation, not exit logging.
US Government Funding & “Honeypot” Fears
- Tor’s origins in US Naval Research and majority US-government funding trigger suspicion for some, who see Tor as a possible “NOBUS” or honeypot.
- Others counter: US agencies also need strong anonymity tools; funding doesn’t prove backdoors, and there’s no visible mass-arrest pattern attributable to breaking Tor’s core design.
Legitimate Uses, Crime, and OpSec
- Use cases cited: bypassing censorship, anonymous journalism/whistleblowing (e.g., SecureDrop), avoiding tracking (e.g., via onion mirrors of major sites), NAT traversal.
- Many arrests of Tor users are attributed in the thread to endpoint compromise, zero-days, metadata (who used Tor when), or blatant operational mistakes, not Tor protocol breaks.
- Consensus: nothing is 100% safe; Tor remains the best widely deployed option for practical anonymity, but users must understand limits and risk tolerance.