ICANN approves domain takedown rules

ICANN’s board of directors has formally approved amendments to its standard registry and registrar contracts aimed at forcing companies to take action against domains involved in DNS abuse.

At its meeting last weekend, the board passed a resolution amending the Registrar Accreditation Agreement and Base gTLD Registry Agreement to include tougher rules on tackling abuse.

Registrars must now “promptly take the appropriate mitigation action(s) that are reasonably necessary to stop, or otherwise disrupt, the Registered Name from being used for DNS Abuse” when provided with evidence of such abuse.

Registries have a similar obligation to take action, but the action might be to refer the abusive domain to the appropriate registrar.

The rules follow the now industry-standard definition of DNS abuse: “malware, botnets, phishing, pharming, and spam (when spam serves as a delivery mechanism for the other forms of DNS Abuse listed)”.

The changes were crafted by ICANN along with registries and registrars and voted through late last year by a hefty majority of both camps.

The two contracts are now in the hands of the ICANN CEO and her lawyers for final action before becoming enforceable.

Registries and registrars vote ‘Yes’ to new DNS abuse rules

ICANN’s contracted registries and registrars have voted to accept new rules requiring them to take action on DNS abuse.

The new rules come after a vote lasting a few months with some quite high thresholds for success.

The current Registrar Accreditation Agreement merely requires registrars to “take reasonable and prompt steps to investigate and respond appropriately to any reports of abuse”, which is pretty vague and barely enforceable.

The amendments, which still need to be rubber-stamped by the ICANN board, make it much clearer what registrars are expected to do in which circumstances. A new paragraph is added that reads:

3.18.2 When Registrar has actionable evidence that a Registered Name sponsored by Registrar is being used for DNS Abuse, Registrar must promptly take the appropriate mitigation action(s) that are reasonably necessary to stop, or otherwise disrupt, the Registered Name from being used for DNS Abuse. Action(s) may vary depending on the circumstances, taking into account the cause and severity of the harm from the DNS Abuse and the possibility of associated collateral damage.

For registries, the new text for the base gTLD Registry Agreement is similar, but with a little more wiggle-room:

Where a Registry Operator reasonably determines, based on actionable evidence, that a registered domain name in the TLD is being used for DNS Abuse, Registry Operator must promptly take the appropriate mitigation action(s) that are reasonably necessary to contribute to stopping, or otherwise disrupting, the domain name from being used for DNS Abuse. Such action(s) shall, at a minimum, include: (i)the referral of the domains being used for the DNS Abuse, along with relevant evidence, to the sponsoring registrar; or (ii) the taking of direct action, by the Registry Operator, where the Registry Operator deems appropriate. Action(s) may vary depending on the circumstances of each case, taking into account the severity of the harm from the DNS Abuse and the possibility of associated collateral damage.

In both cases, DNS abuse is defined by the now industry standard line: “malware, botnets, phishing, pharming, and spam (when spam serves as a delivery mechanism for the other forms of DNS Abuse listed in this Section)”.

There are a few other quality of life updates, such as the requirement for registrars to acknowledge receipt of abuse reports and to have their abuse reporting mechanism “conspicuously and readily accessible from” their home pages.

ICANN needed registrars representing over 90% of registered gTLD domains (adjusted slightly to make GoDaddy’s voice less powerful). That threshold was passed last week, with 94% of domains voting in favor of the amendments.

For registries, ICANN required a simple majority of registries (counted by contract rather than company) and for all registries voting in favor to have been responsible for two thirds of all registry fees paid last year.

Judging by the financial thresholds, .com and .net, which are not on the base RA, were not involved.

Did I find a murder weapon in a zone file?

Registrars are usually very reluctant to police the content of web sites by taking down domains they manage, but they quite often make an exception when the web site in question calls for violence. But what if the site itself attempts to physically harm visitors through their screens?

It sounds a bit mad, but I think I’ve found such a site.

I recently randomly came across a domain name that caught my eye while scrolling through a zone file. I’m not going to reveal the domain here, but it consisted of three words across the dot and could be taken as an instruction to “murder” a specific, but unnamed, individual.

Expecting humor, I visited the domain out of curiosity and was confronted by a blank page that rapidly flashed between two background colors, creating a strobe effect. There was no other content.

My first impression was that the site had been created in order to trigger seizures in photosensitive epileptics. The CSS seemed to confirm that the strobe effect fell within the frequency range that the charity Epilepsy Action says can cause such seizures.

This raised an interesting question: could this be considered “DNS abuse”?

The DNS Abuse Institute’s definition (pdf) says DNS Abuse consists of “malware, botnets, phishing, pharming, and spam (when it serves as a delivery mechanism for the other forms of DNS Abuse)”.

DNSAI says registries and registrars “must” act on these five categories of abuse, but it adds that there are some categories of web content where registrars “should” take action. Its Framework to Address Abuse, which has been endorsed by dozens of registries and registrars, states:

Specifically, even without a court order, we believe a registry or registrar should act to disrupt the following forms of Website Content Abuse: (1) child sexual abuse materials (“CSAM”); (2) illegal distribution of opioids online; (3) human trafficking; and (4) specific and credible incitements to violence. Underlying these Website Content Abuses is the physical and often irreversible threat to human life.

Epileptic seizures can be fatal. A school friend of mine did not make it out of his teens due to one. Even when non-fatal, they are dangerous and clearly unpleasant.

So if a site encouraging physical violence “should” be taken down, what about a site that seems designed to actively physically attack individuals, no incitement required? That’s a reasonable question, right?

I filed an abuse report with the registrar managing the domain and was told it did not violate its acceptable use policies.

Attacking epileptics with flashing images sent online has been a criminal offence in the UK since October 26, when the controversial Online Safety Act 2023 was enacted.

A component of the Act is named Zach’s Law, after an eight-year-old boy who in 2020 was attacked with flashing images by internet wankers after he carried out a sponsored walk for the Epilepsy Society.

The Act makes it illegal to send a flashing image to somebody you know is epileptic with the intent to harm them. You can get up to five years imprisonment and a fine.

Is ICANN toothless in the face of DNS abuse?

Concerns have been raised that ICANN may lack the tools to tackle DNS abuse using its contracts with registries and registrars.

The new report from the GNSO’s “small team” on abuse has highlighted two “gaps” in the current Compliance regime that may be allowing registrars to get away with turning a blind eye to abusive customers.

The current version of the standard Registrar Accreditation Agreement calls for registrars to maintain an abuse contact email and to “take reasonable and prompt steps to investigate and respond appropriately to any reports of abuse.”

The problem, the small team report finds, is that ICANN Compliance doesn’t seem to have a standard definition of “reasonable”, “prompt”, and “appropriately”. The contract doesn’t require any specific remediations from the registrar.

“Members of the small team are concerned that this interpretation may allow DNS abuse to remain unmitigated, depending upon the registrar’s specific domain name use and abuse policies,” the report states.

Judging by conversations at ICANN 75 last month, it’s apparently the first time Compliance has gone on the record about how it enforces this part of the contract.

It’s quite rare for ICANN to issue a public breach notice to a registrar over its failure to respond to abuse reports and when it does, it tends to relate to the registrar’s failure to keep records showing how it responded.

I can’t find any instances where Compliance has canned a registrar for allowing abusive domains — typically defined as those hosting malware, phishing, botnets, pharming and some spam — to remain active after an abuse report.

The small team’s report also thinks there’s a blind spot in ICANN’s standard Registry Agreement, which in turn requires registries to include, in their Registry-Registrar Agreements, provisions requiring anti-abuse terms in the registrars’ Registration Agreements.

This complex chain of contractual provisions doesn’t seem to be enforced, the small team notes, saying “further consideration may need to be given to what Registries are doing to ensure the text is indeed included in the Registration Agreement (ie Registries enforcing their own Registry-Registrar Agreements”.

The small team recommends that contracted parties talk further with ICANN about possible contract changes or best practices documents before going ahead with policy-making. The GNSO Council will address the recommendations later this month.

ICANN to mull bulk registration ban

ICANN policymakers are to take a look at banning bulk domain registrations in ongoing efforts to combat DNS abuse.

While in the very early stages of discussion, the GNSO Council is being urged to start gathering data “to further explore the role that bulk registrations play in DNS Abuse” and “to consider whether further action on bulk registrations is deemed necessary”

The recommendation is among several in a newly published report of a cross-constituency GNSO “small team”, which may lead to “tightly focused and scoped policy development”.

While acknowledging “there are also examples in which bulk registrations are used for legitimate purposes”, the report states:

The small team recommends that the GNSO Council requests the Registrar Stakeholder Group and others (for example, ICANN org, the RySG and the DNSAI) to further explore the role that bulk registrations play in DNS Abuse as well as measures that Registrars may have already put in place to address this vector. Based on the feedback received, the GNSO Council will consider whether further action on bulk registrations is deemed necessary.

The report is to be considered later this month at the GNSO Council’s monthly meeting. Any actual policy outcome, if any, will be years away.

NetBeacon goes live for DNS abuse reporting

The DNS Abuse Institute has gone live with its new clearinghouse for DNS abuse reports, NetBeacon.

The service allows anyone to report any domain for four types of abuse — malware, phishing, botnets and spam — and any registry or registrar can sign up to receive the reports in a normalized feed via email or API.

The idea is to make it easier for domain companies to act on reports of abusive customers, as DNSAI director Graeme Bunton told us a few months ago.

NetBeacon is free for both reporters and registrars and is being funded by .org manager Public Interest Registry.

Some of the technology underpinning the service is being provided by CleanDNS.

ICANN extends Covid-19 abuse monitoring to Ukraine war

ICANN has started monitoring domains related to the war in Ukraine for potential abuse, expanding an ongoing project related to the Covid-19 pandemic.

CEO Göran Marby has during multiple sessions at ICANN 73 this week said that the Org will soon announce an extension of its DNSTICR project — pronounced “DNS Ticker” and standing for Domain Name Security Threat Information Collection & Reporting.

The plan is to alert registrars about Ukraine-related domain names being used to scam people or drop malware.

“There will be coming up more information about this very soon, but we have decided to also add names in relationship to the conflict in Ukraine,” Marby said during a session with the Commercial Stakeholders Group.

DNSTICR was launched in March 2020, when the pandemic was in full swing, to find new domains containing keywords such as “covid”, “pandemic” and “coronavirus”, and check them against domain abuse lists.

From May 2020 to August last year, it flagged 210,939 pandemic-related domains, and found that 3,791 of them were malicious with “high confidence”.

CTO John Crain said in a session on Monday: “There’s a lot of stuff in the press and some technical papers out there that show clearly that the bad guys, as always, have, once again, pivoted to whatever is happening in the world. So if we can do a little bit to help, we will.”

PIR to offer industry FREE domain abuse clearinghouse

The DNS Abuse Institute will soon launch a free service designed to make it easier to report abuse and for registries and registrars to act upon it.

The Institute, which is funded by .org manager Public Interest Registry, is working on a system provisionally called CART, for Centralized Abuse Reporting Tool, an ambitious project that would act as a clearinghouse for abuse reports across the industry.

The plan is to offer the service for free to reporters and registrars alike, with a beta being offered to registrars late next month and a public launch hopefully before ICANN 74 in June.

DNSAI director Graeme Bunton said that CART is meant to solve the “mess” of current abuse reporting systems.

For abuse reporters, the idea is to give them a one-stop shop for their reports, across all gTLDs and registrars. CART would take their complaints, normalize them, furnish them with additional information from sources such as Whois records and domain block-lists, and shunt them off to the registrar of record.

“Registrars get boatloads of abuse reports every day,” Bunton said. “Hundreds to thousands. They’re often duplicative, often unevidenced — almost always. There’s no standardization. So they’re having to spend a lot of time reading and parsing these abuse reports.”

“They’re spending a huge amount of time triaging tickets that don’t make the internet any better,” he said. “It felt like trying to solve this problem across every individual registry and registrar was not going to work, and that a centralizing function that sits in the middle and absorbs a lot of the complexity would make a real difference, and we’ve been working towards that.”

CART reporters would be authenticated, and their reports would be filed through forms that normalized the data to make them easier for registrars to understand. There will be “evidence requirements” to submit a report.

“It’s a common lament that the abuse@ email that registrars have to publish are filled with garbage,” Bunton said. “This is intended to clean that up, as well as make it easier for reporters.”

Registrars will be able to white-label these forms on their own sites, replacing or adding to existing reporting mechanisms, which will hopefully drive adoption of the tool, Bunton said.

Registrars will be able to use an API to pull the abuse feed into their existing ticketing workflows, or simply receive the reports via email.

The plan is to send these enhanced reports to registrars’ publicly listed abuse@ addresses, whether they opt into the CART system or not, Bunton said.

One feature idea — possibly in a version 2 release — is to have a reputation-scoring function in which registrars can flag reporters as reliable, facilitating on-the-fly “trusted notifier” relationships.

While the DNSAI is focusing to the industry definition of “DNS abuse” — phishing, pharming, malware, botnets and a subset of spam — the plan is to not limit reporters to just those categories.

Copyright infringement claims, for example, would be acceptable forms of abuse report, if the registrar enables that option when they embed the CART forms on their own sites.

CART will most likely be renamed to something with “better mass-market appeal” before it launches, Bunton said, but there will be no charge to reporters or registrars.

“This is all free, with no plans to do cost-recovery or anything like that,” he said.

While Bunton didn’t want to comment, I think it’s unlikely that these projects would be going ahead, at least not for free, had PIR been turned into a for-profit company under its proposed acquisition by Ethos Capital, which was blocked by ICANN a couple of years ago.

A second project DNSAI is working on is called Intelligence.

This will be somewhat similar to ICANN’s own Domain Abuse Activity Reporting (DAAR) system, but with greater granularity, such as giving the ability to see abuse trends by registry or registrar.

The current plan is to have a preview of Intelligence available in June, with a launch in July.

Verisign and PIR join new DNS abuse group

The domain name industry has just got its fourth (by my count) DNS abuse initiative, with plans for work on “trusted notifier” programs and Public Interest Registry and Verisign as members.

topDNS, which announced itself this week, is a project out of eco, the German internet industry association. It said its goals are:

the exchange of best practices, the standardisation of abuse reports, the development of a trusted notifier framework, and awareness campaigns towards policy makers, decision-makers and expert groups

eco’s Thomas Rickert told DI that members inside and outside the industry had asked for such an initiative to combat “the narrative that industry is not doing enough against an ever-increasing problem”.

He said there’s a “worrying trend” of the domain industry being increasingly seen as an easy bottleneck to get unwelcome content taken down, rather than going after the content or hosting provider.

“There is not an agreed-upon definition of what constitutes DNS abuse,” he said.

“There are groups interested in defining DNS abuse very broadly, because it’s more convenient for them I guess to go to a registrar or registry and ask for a domain takedown rather than trying to get content taken down with a hosting company,” he said.

topDNS has no plans to change the definition of “DNS abuse” that has already been broadly agreed upon by the legit end of the industry.

The DNS Abuse Framework, which was signed by 11 major registries and registrars (now, it’s up to 48 companies) in 2019 defines it as “malware, botnets, phishing, pharming, and spam (when it serves as a delivery mechanism for the other forms of DNS Abuse)”.

This is pretty much in line with their ICANN contractual obligations; ICANN itself shudders away from being seen as a content regulator.

The big asterisk next to “spam” perhaps delineates “domains” from “content”, but the Framework also recommends that registries and registrars should act against content when it comprises child sexual abuse material, illegal opioid sales, human trafficking, and “specific and credible” incitements to violence.

Rickert said the plan with topDNS is to help “operationalize” these definitions, providing the domain industry with things like best practice documents.

Of particular interest, and perhaps a point of friction with other parties in the ecosystem in future, is the plan to work on “the development of a trusted notifier framework”.

Trusted notifier systems are in place at a handful of gTLD and ccTLD registries already. They allow organizations — typically law enforcement or Big Content — a streamlined, structured path to get domains taken down when the content they lead to appears to be illegal.

The notifiers get a more reliable outcome, while the registries get some assurances that the notifiers won’t take the piss with overly broad or spammy takedown requests.

topDNS will work on templates for such arrangements, not on the arrangements themselves, Rickert said. Don’t expect the project to start endorsing certain notifiers.

Critics such as the Electronic Frontier Foundation find such programs bordering on censorship and therefore dangerous to free speech.

While the topDNS initiative only has six named members right now, it does have Verisign (.com and .net) and PIR (.org), which together look after about half of all extant domains across all TLDs. It also has CentralNic, a major registrar group and provider of back-end services for some of the largest new gTLDs.

“Verisign is pleased to support the new topDNS initiative, which will help bring together stakeholders with an interest in combating and mitigating DNS security threats,” a company spokesperson said.

Unlike CentralNic and PIR, Verisign is not currently one of the 48 signatories of the DNS Abuse Framework, but the spokesperson said topDNS is “largely consistent” with that effort.

Verisign has also expressed support for early-stage trusted notifier framework discussions being undertaken by ICANN’s registry and registrar stakeholder groups.

PIR also has its own separate project, the DNS Abuse Institute, which is working on similar stuff, along with some tools to support the paperwork.

DNSAI director Graeme Bunton said: “I see these efforts as complementary, not competing, and we are happy to support and participate in each of them.” He’s going to be on topDNS’s inaugural Advisory Council, he and Rickert said.

Rickert and Bunton both pointed out that topDNS is not going to be limited to DNS abuse issues alone — that’s simply the most pressing current matter.

Rickert said issues such as DNS over HTTP and blockchain naming systems could be of future interest.

Most registrars fail ICANN abuse audit

The large majority of accredited registrars failed an abuse-related audit at the first pass, according to ICANN.

(UPDATE October 14, 2021: ICANN disagrees with this characterization.)

The audit of 126 registrars, representing over 90% of all registered gTLD domains, founds that 111 were “not fully compliant with the [Registrar Accreditation Agreement’s] requirements related to the receiving and handling of DNS abuse reports”.

Only 15 companies passed with flying colors, ICANN said.

A further 92 have already put in place changes to address the identified concerns, with 19 more still struggling to come into compliance.

The particular parts of the RAA being audited require registrars to publish an abuse email address that it monitored 24/7 and to take action on well-founded cases of abuse within 24 hours of notification.

The results of the audit, carried out by ICANN Compliance and KPMG, can be found here (pdf).