12/10 2017
1 - Internet Chemotherapy
Internet Chemotherapy was a 13 month project between Nov 2016 - Dec 2017. It has been known under names such as ‘BrickerBot’, ‘bad firmware upgrade’, ‘ransomware’, ‘large-scale network failure’ and even ‘unprecedented terrorist actions.’ That last one was a little harsh, Fernandez, but I guess I can’t please everybody.
You can download the module which executes the http and telnet-based payloads from this router at http://91.215.104.140/mod_plaintext.py. Due to platform limitations the module is obfuscated single threaded python, but the payloads are in plain view and should be easy to figure out for any programmer worth his/her/hir salt. Take a look at the number of payloads, 0-days and techniques and let the reality sink in for a moment. Then imagine what would’ve happened to the Internet in 2017 if I had been a blackhat dedicated to building a massive DDoS cannon for blackmailing the biggest providers and companies. I could’ve disrupted them all and caused extraordinary damage to the Internet in the process.
My ssh crawler is too dangerous to publish. It contains various levels of automation for the purpose of moving laterally through poorly designed ISP networks and taking them over through only a single breached router. My ability to commandeer and secure hundreds of thousands of ISP routers was the foundation of my anti-IoT botnet project as it gave me great visibility of what was happening on the Internet and it gave me an endless supply of nodes for hacking back. I began my non-destructive ISP network cleanup project in 2015 and by the time Mirai came around I was in a good position to react. The decision to willfully sabotage other people’s equipment was nonetheless a difficult one to make, but the colossally dangerous CVE-2016-10372 situation ultimately left me with no other choice. From that moment on I was all-in.
I am now here to warn you that what I’ve done was only a temporary band-aid and it’s not going to be enough to save the Internet in the future. The bad guys are getting more sophisticated, the number of potentially vulnerable devices keep increasing, and it’s only a matter of time before a large scale Internet-disrupting event will occur. If you are willing to believe that I’ve disabled over 10 million vulnerable devices over the 13-month span of the project then it’s not far-fetched to say that such a destructive event could’ve already happened in 2017.
YOU SHOULD WAKE UP TO THE FACT THAT THE INTERNET IS ONLY ONE OR TWO SERIOUS IOT EXPLOITS AWAY FROM BEING SEVERELY DISRUPTED. The damage of such an event is immeasurable given how digitally connected our societies have become, yet CERTs, ISPs and governments are not taking the gravity of the situation seriously enough. ISPs keep deploying devices with exposed control ports and although these are trivially found using services like Shodan the national CERTs don’t seem to care. A lot of countries don’t even have CERTs. Many of the world’s biggest ISPs do not have any actual security know-how in-house, and are instead relying on foreign vendors for help in case anything goes wrong. I’ve watched large ISPs withering for months under conditioning from my botnet without them being able to fully mitigate the vulnerabilities (good examples are BSNL, Telkom ZA, PLDT, from time to time PT Telkom, and pretty much most large ISPs south of the border). Just look at how slow and ineffective Telkom ZA was in dealing with its Aztech modem problem and you will begin to understand the hopelessness of the current situation. In 99% of the problem cases the solution would have simply been for the ISPs to deploy sane ACLs and CPE segmentation, yet months later their technical staff still hasn’t figured this out. If ISPs are unable to mitigate weeks and months of continuous deliberate sabotage of their equipment then what hope is there that they would notice and fix a Mirai problem on their networks? Many of the world’s biggest ISPs are catastrophically negligent and this is the biggest danger by a landslide, yet paradoxically it should also be the easiest problem to fix.
I’ve done my part to try to buy the Internet some time, but I’ve gone as far as I can. Now it’s up to you. Even small actions are important. Among the things you can do are:
-
Review your own ISP’s security through services such as Shodan and take them to task over exposed telnet, http, httpd, ssh, tr069 etc. ports on their networks. Refer them to this document if you have to. There’s no good reason why any of these control ports should ever be accessible from the outside world. Exposing control ports is an amateur mistake. If enough customers complain they might actually do something about it!
-
Vote with your wallet! Refuse to buy or use ‘intelligent’ products unless the manufacturer can prove that the product can and will receive timely security updates. Find out about the vendor’s security track record before giving them your hard-earned money. Be willing to pay a little bit more for credible security.
-
Lobby your local politicians and government officials for improved security legislation for IoT (Internet of Things) devices such as routers, IP cameras and ‘intelligent’ devices. Private or public companies currently lack the incentives for solving this problem in the immediate term. This matter is as important as minimum safety requirements for cars and general electrical appliances.
-
Consider volunteering your time or other resources to underappreciated whitehat organizations such as GDI Foundation or Shadowserver Foundation. These organizations and people make a big difference and they can significantly amplify the impact of your skillset in helping the Internet.
-
Last but not least, consider the long-shot potential of getting IoT devices designated as an ‘attractive nuisance’ through precedent-setting legal action. If a home owner can be held liable for burglar/trespasser getting injured then I don’t see why a device owner (or ISP or manufacturer) shouldn’t be held liable for the damage that was caused by their dangerous devices being exploitable through the Internet. Attribution won’t be a problem for Layer 7 attacks. If any large ISPs with deep pockets aren’t willing to fund such precedent cases (and they might not since they fear that such precedents could come back to haunt them) we could even crowdfund such initiatives over here and in the EU. ISPs: consider your volumetric DDoS bandwidth cost savings in 2017 as my indirect funding of this cause and as evidence for its potential upside.
2 - Timeline
Here are some of the more memorable events of the project:
-
Deutsche Telekom Mirai disruption in late November 2016. My hastily assembled initial TR069/64 payload only performed a ‘route del default’ but this was enough to get the ISP’s attention to the problem and the resulting headlines alerted other ISPs around the world to the unfolding disaster.
-
Around January 11-12 some Mirai-infected DVRs with exposed control port 6789 ended up getting bricked in Washington DC, and this made numerousheadlines. Gold star to Vemulapalli for determining that Mirai combined with /dev/urandom had to be ‘highly sophisticated ransomware’. Whatever happened to those 2 unlucky souls in Europe?
-
In late January 2017 the first genuine large-scale ISP takedown occured when Rogers Canada’s supplier Hitron carelessly pushed out new firmware with an unauthenticated root shell listening on port 2323 (presumably this was a debugging interface that they forgot to disable). This epic blunder was quickly discovered by Mirai botnets, and the end-result was a large number of bricked units.
-
In February 2017 I noticed the first Mirai evolution of the year, with both Netcore/Netis and Broadcom CLI-based modems being attacked. The BCM CLI would turn out to become one of the main Mirai battlegrounds of 2017, with both the blackhats and me chasing the massive long tail of ISP and model-specific default credentials for the rest of the year. The ‘broadcom’ payloads in the above source may look strange but they’re statistically the most likely sequences to disable any of the endless number of buggy BCM CLI firmwares out there.
-
In March 2017 I significantly increased my botnet’s node count and started to add more web payloads in response to the threats from IoT botnets such as Imeij, Amnesia and Persirai. The large-scale takedown of these hacked devices created a new set of concerns. For example, among the leaked credentials of the Avtech and Wificam devices there were logins which strongly implied airports and other important facilities, and around April 1 2017 the UK government officials warned of a ‘credible cyber threat’ to airports and nuclear facilities from ‘hacktivists.’ Oops.
-
The more aggressive scanning also didn’t escape the attention of civilian security researchers, and in April 6 2017 security company Radware published an article about my project. The company trademarked it under the name ‘BrickerBot.’ It became clear that if I were to continue increasing the scale of my IoT counteroffensive I had to come up with better network mapping/detection methods for honeypots and other risky targets.
-
Around April 11th 2017 something very unusual happened. At first it started like so many other ISP takedowns, with a semi-local ISP called Sierra Tel running exposed Zyxel devices with the default telnet login of supervisor/zyad1234. A Mirai runner discovered the exposed devices and my botnet followed soon after, and yet another clash in the epic BCM CLI war of 2017 took place. This battle didn’t last long. It would’ve been just like any of the hundreds of other ISP takedowns in 2017 were it not for something very unusual occuring right after the smoke settled. Amazingly, the ISP didn’t try to cover up the outage as some kind of network issue, power spike or a bad firmware upgrade. They didn’t lie to their customers at all. Instead, they promptly published a press release about their modems having been vulnerable which allowed their customers to assess their potential risk exposure. What did the most honest ISP in the world get for its laudable transparency? Sadly it got little more than criticism and bad press. It’s still the most depressing case of ‘why we can’t have nice things’ to me, and probably the main reason for why 99% of security mistakes get covered up and the actual victims get left in the dark. Too often ‘responsible disclosure’ simply becomes a euphemism for ‘coverup.’
-
On April 14 2017 DHS warned of ‘BrickerBot Threat to Internet of Things’ and the thought of my own government labeling me as a cyber threat felt unfair and myopic. Surely the ISPs that run dangerously insecure network deployments and the IoT manufacturers that peddle amateurish security implementations should have been fingered as the actual threat to Americans rather than me? If it hadn’t been for me millions of us would still be doing their banking and other sensitive transactions over hacked equipment and networks. If anybody from DHS ever reads this I urge you to reconsider what protecting the homeland and its citizens actually means.
-
In late April 2017 I spent some time on improving my TR069/64 attack methods, and in early May 2017 a company called Wordfence (now Defiant) reported a significant decline in a TR069-exploiting botnet that had previously posed a threat to Wordpress installations. It’s noteworthy that the same botnet temporarily returned a few weeks later using a different exploit (but this was also eventually mitigated).
-
In May 2017 hosting company Akamai reported in its Q1 2017 State of the Internet report an 89% decrease in large (over 100 Gbps) DDoS attacks compared with Q1 2016, and a 30% decrease in total DDoS attacks. The largest attack of Q1 2017 was 120 Gbps vs 517 Gbps in Q4 2016. As large volumetric DDoS was one of the primary signatures of Mirai this felt like concrete justification for all the months of hard work in the IoT trenches.
-
During the summer I kept improving my exploit arsenal, and in late July I performed some test runs against APNIC ISPs. The results were quite surprising. Among other outcomes a few hundred thousand BSNL and MTNL modems were disabled and this outage become headline news in India. Given the elevated geopolitical tensions between India and China at the time I felt that there was a credible risk of the large takedown being blamed on China so I made the rare decision to publically take credit for it. Catalin, I’m very sorry for the abrupt ‘2 day vacation’ that you had to take after reporting the news.
-
Previously having worked on APNIC and AfriNIC, on August 9th 2017 I also launched a large scale cleanup of LACNIC space which caused problems for various providers across the subcontinent. The attack made headlines in Venezuela after a few million cell phone users of Movilnet lost service. Although I’m personally against government surveillance of the Internet the case of Venezuela is noteworthy. Many of the LACNIC ISPs and networks have been languishing for months under persistent conditioning from my botnet, but Venezuelan providers have been quick to fortify their networks and secure their infrastructure. I believe this is due to Venezuela engaging in far more invasive deep packet inspection than the other LACNIC countries. Food for thought.
-
In August 2017 F5 Labs released a report called “The Hunt for IoT: The Rise of Thingbots” in which the researchers were perplexed over the recent lull in telnet activity. The researchers speculated that the lack of activity may be evidence that one or more very large cyber weapons are being built (which I guess was in fact true). This pieceis to my knowledge the most accurate assessment of the scope of my project but fascinatingly the researchers were unable to put two and two together in spite of gathering all the relevant clues on a single page.
-
In August 2017 Akamai’s Q2 2017 State of the Internet report announces the first quarter in 3 years without the provider observing a single large (over 100 Gbps) attack, and a 28% decrease in total DDoS attacks vs Q1 2017. This seems like further validation of the cleanup effort. This phenomenally good news is completely ignored by the mainstream media which operates under an ‘if it bleeds it leads’ mentality even when it comes to information security. This is yet another reason why we can’t have nice things.
-
After the publication of CVE-2017-7921 and 7923 in September 2017 I decided to take a closer look at Hikvision devices, and to my horror I realized that there’s a technique for botting most of the vulnerable firmwares that the blackhats hadn’t discovered yet. As a result I launched a global cleanup initiative around mid-September. Over a million DVRs and cameras (mainly Hikvision and Dahua) were disabled over a span of 3 weeks and publications such as IPVM.com wrote several articles about the attacks. Dahua and Hikvision wrote press releasesmentioning or alluding to the attacks. A huge number of devices finally got their firmwares upgraded. Seeing the confusion that the cleanup effort caused I decided to write a quick summary for the CCTV people at http://depastedihrn3jtw.onion.link/show.php?md5=62d1d87f67a8bf485d43a05ec32b1e6f (sorry for the NSFW language of the pastebin service). The staggering number of vulnerable units that were online months after critical security patches were available should be the ultimate wakeup call to everyone about the utter dysfunctionality of the current IoT patching process.
-
Around September 28 2017 Verisign releases a report saying that DDoS attacks declined 55% in Q2 2017 vs Q1, with a massive 81% attack peak decline.
-
On November 23rd 2017 the CDN provider Cloudflare reports that ‘in recent months, Cloudflare has seen a dramatic reduction in simple attempts to flood our network with junk traffic.’ Cloudflare speculates it could’ve partly been due to their change in policies, but the reductions also line up well with the IoT cleanup activities.
-
At the end of November 2017 Akamai’s Q3 2017 State of the Internet report sees a small 8% increase in total DDoS attacks for the quarter. Although this was a significant reduction compared to Q3 2016 the slight uptick serves as a reminder of the continued risks and dangers.
-
As a further reminder of the dangers a new Mirai strain dubbed ‘Satori’ reared its head in November-December of 2017. It’s particularly noteworthy how quickly the botnet managed to grow based on a single 0-day exploit. This event underlines the current perilous operating state of the Internet, and why we’re only one or two severe IoT exploits away from widespread disruption. What will happen when nobody is around to disable the next threat? Sinkholing and other whitehat/‘legal’ mitigations won’t be enough in 2018 just like they weren’t enough in 2016. Perhaps in the future governments will be able to collaborate on a counterhacking task force with a global mandate for disabling particularly severe existential threats to the Internet, but I’m not holding my breath.
-
Late in the year there were also some hysterical headlines regarding a new botnet that was dubbed ‘Reaper’ and ‘IoTroop’. I know some of you will eventually ridicule those who estimated its size at 1-2 million but you should understand that security researchers have very limited knowledge of what’s happening on networks and hardware that they don’t control. In practice the researchers could not possibly have known or even assumed that most of the vulnerable device pool had already been disabled by the time the botnet emerged. Give the ‘Reaper’ one or two new unmitigated 0-days and it’ll become as terrifying as our worst fears.
3 - Parting Thoughts
I’m sorry to leave you in these circumstances, but the threat to my own safety is becoming too great to continue. I have made many enemies. If you want to help look at the list of action items further up. Good luck.
There will also be those who will criticize me and say that I’ve acted irresponsibly, but that’s completely missing the point. The real point is that if somebody like me with no previous hacking background was able to do what I did, then somebody better than me could’ve done far worse things to the Internet in 2017. I’m not the problem and I’m not here to play by anyone’s contrived rules. I’m only the messenger. The sooner you realize this the better.
-Dr Cyborkian a.k.a. janit0r, conditioner of ‘terminally ill’ devices.