ORGANISATIONS LEAST
+46%
CRITICAL FINDING
+11%
ORGANISATIONS ENFORCED
+60%
FRENCH COMPANIES
+49%
Imagine you walked down a street of 129 houses and checked — without touching a single door — which ones had their keys dangling from the lock, their garage doors visibly up, and their alarm codes taped to the mailbox. That's what this paper does for the French maritime sector's internet-facing attack surface, and the results are roughly what you'd expect if nobody on the block had ever heard of a locksmith. The committed claim: a first systematic, passive measurement of external cyber exposure across France's NIS2-designated maritime sector — 103 French companies and 26 international operators — using only publicly available data (DNS, DMARC, TLS certs, Shodan, cached pages, cloud bucket naming). No active probing, no exploitation. The March 2026 scan surfaced 475 verified findings across those 129 organisations, averaging 3.7 per org. 46% had at least one high or critical finding. 11% had at least one critical finding — exposed databases, unpatched services with known CVEs, a credentials file sitting in public, listable cloud buckets, and a subdomain ripe for takeover. The single most common gap: 60% of organisations had no enforced DMARC policy, meaning anyone could spoof their email domain. French companies performed measurably worse than international operators active in French waters: 49% vs 35% at high-or-critical risk, and 4.0 vs 2.3 findings per organisation. This is the ladder position that matters — it's not France vs. some abstract ideal, it's France vs. multinational operators subject to comparable regulation but apparently running tighter ships. The gap is roughly 40% worse exposure rate and nearly double the finding density. No prior public study gives this comparison for the French maritime sector specifically. The architecture here is methodological, not algorithmic: passive OSINT reconnaissance using a well-known toolkit (DNS enumeration, DMARC parsing, Shodan queries, certificate transparency logs, search-engine dorking, cloud-bucket naming heuristics). The key structural choice is strict passivity — no connection attempts beyond ordinary public web requests. This is not a penetration test; it's a measurement study. That passivity is both the paper's integrity strength (no legal grey zone, no ethical ambiguity) and its limitation (can't assess anything behind a login, can't confirm exploitability). The most striking result is the disclosure experiment. Between March 2 and 18, the authors sent individual reports to 43 of the audited domains. When they rescanned all 123 reachable organisations in November 2026, 52% of critical/high findings were gone for notified orgs and 54% were gone for unnotified orgs. Private responsible disclosure had literally no measurable effect. Remediation happened at the same rate regardless. This is an important negative result: it suggests that whatever drives patching in this sector, it isn't a polite email from a researcher. Possible explanations include automated infrastructure rotation, hosting-provider-driven updates, or simply that the emails were ignored and the same background processes fixed things everywhere. The integrity regime is straightforward for a measurement study. Everything is passive and reproducible; the data is anonymised and aggregated; Zenodo deposit provides a persistent record. The main weakness is sample construction — the 129 orgs were selected by the authors, and we don't know the full universe of NIS2-designated maritime entities in France. There's no pre-registration, but for a descriptive measurement study that's less of a concern than it would be for a hypothesis-testing paper. The 475 findings were manually verified, which is good, but introduces human judgment into what counts as 'verified.' The milestone question for this line of work is whether passive exposure measurement can become a routine, sector-wide monitoring tool under NIS2. Right now we have a single snapshot (March 2026) and one follow-up (November 2026). The next concrete number to watch is whether a continuous monitoring cadence — quarterly or monthly — reveals a remediation half-life for different finding categories. If the median critical finding persists for, say, 6 months regardless of disclosure, that's a policy-relevant number that could drive regulatory enforcement. The obvious experiment not run: active scanning. The authors stayed strictly passive, which is legally clean but means they can't assess actual exploitability, internal network exposure, or the severity of findings behind authentication. The honest read is (a) — legal and ethical constraints, not compute. Active scanning of 129 organisations without explicit authorisation would be legally problematic in France. The second missing experiment is a controlled disclosure trial with proper randomisation and larger N; the current 43-domain disclosure was done ad hoc between March 2-18, not as a randomised controlled trial. That's likely a matter of scope and the paper being a first pass, not a hidden negative result.
The analogy
You walk down a street checking which houses have keys in the lock, garage doors up, and alarm codes on the mailbox — without touching anything. Then you leave polite notes for half the homeowners. Eight months later, the same fraction of doors are locked regardless of whether they got your note.
Neighbourhood walk
Checking visible security gaps (open garages, keys in locks) from the sidewalk without touching property
Passive OSINT scan
Checking internet-facing exposure (DNS, DMARC, Shodan, certs) without active probing or login attempts
Honest assessment
A clean, modest measurement study that establishes a first baseline for a sector that had none publicly. The disclosure null result is the most interesting finding — it challenges the assumption that responsible disclosure drives remediation. The methods are standard OSINT, the sample is small and author-selected, and the paper makes no overclaims. Useful primarily for NIS2 policymakers and French maritime CERTs who need numbers to justify enforcement, less so for security researchers looking for methodological novelty.
What Is CLAIMS?
57
/ 100
BedrockNews applies a structured analytical framework called CLAIMS to academic papers and technical research releases.
How CLAIMS worksClaim
3/5
Ladder
2/5
Architecture
2/5
Integrity
3/5
Milestone
3/5
Successor
4/5
What is this paper actually committing to?
Genuine first for French maritime specifically: no prior public passive measurement of this NIS2 sector exists. The disclosure null result is the sharper contribution — polite emails don't move the needle.
The field fight
Does responsible disclosure to individual organisations improve cybersecurity posture, or does only regulatory enforcement / systemic infrastructure change move the needle?
Side A
Responsible disclosure works — informed defenders patch faster
Held by: CERT coordination models, bug bounty platforms, ISO 29147
Side B
Disclosure is theatre — remediation is driven by infrastructure churn, vendor patching, and regulatory pressure, not researcher emails
Held by: This paper, large-scale measurement studies showing patching unrelated to notification
Milestone tracker
Unlocks: Sector-wide remediation half-life metrics that could trigger NIS2 enforcement actions
Current
129
orgs × 2 scans
this paper
Next
500+
orgs × quarterly
continuous NIS2 monitoring
~1-2 years if adopted by ANSSI or a French maritime CERT
Verified Facts (8)
129 organisations assessed: 103 French companies and 26 international operators active in French waters
475 findings retained after manual verification, 3.7 per organisation
46% of organisations had at least one high or critical finding and 11% at least one critical finding
60% of organisations had no enforced DMARC policy
French companies: 49% at high or critical risk, 4.0 findings per org; international operators: 35%, 2.3 findings per org
Individual reports sent for 43 audited domains between 2 and 18 March 2026
Rescan of 123 organisations in November 2026: 52% of critical/high findings gone for notified orgs, 54% for unnotified orgs
5 pages, 5 tables; also deposited on Zenodo, doi:10.5281/zenodo.24279346
Practical Implications (4)
Maritime CERTs and NIS2 regulators now have a citable baseline: 46% of French maritime orgs have high/critical external exposure, and responsible disclosure alone doesn't fix it
DMARC enforcement should be a low-hanging regulatory mandate — 60% non-enforcement is an email-spoofing gift to attackers
Any org can replicate this passive methodology against their own sector using the same public tools (DNS, Shodan, cert transparency) with no legal risk
The disclosure null result suggests sector-wide mandates or automated infrastructure updates are more effective than individual notification
More latest news
Read all