Every problem hides opportunities! I can see a future where documentation will be replaced by a plugin to an AI code service. Instead of providing users with documentation on how to use the package, devs will be training an LLM on how to assist the user in generating the interface code. An elaborate Chat GPT prompt for instance.
If the tool exists and has minimal overhead, I don't think it is a matter of permission but a matter of necessity. CBMC adds about 30% overhead over just unit testing, in my experience. It does require a different idiomatic programming style to use correctly, and there is a learning curve. Some intuition is required to learn how to use the tool effectively and efficiently, but this can be taught.
For system programming or firmware, it's incredibly useful. I've used this tool in conjunction with threat modeling and attack surface minimization to significantly improve the safety and security of software written in C. At this point, I would not consider working on any C project that did not make use of this tool.
Not fun, a serious tool, and extremely easy to use, unlike other formal methods. It works with your source code, not on some rewritten abstraction of it.
Embedded, automotive, space use it regularly.
I've setup cbmc and goto-analyzer for string searching algos here https://github.com/rurban/smart and really found some bugs in old published algos.
Assuming that you mean working on cool research. You can interview and come in as a DARPA Program manager where you need to bring in a grand vision and would be in charge of revolutionizing a field(s). But what is it that you would do day to day? A DARPA PM doesn't do actual research but instead has access to the network of the best minds in research and uses them to shape a multi-million 5 yr 'program' and arrange for funding and transition paths to DoD and private industry. A PM's tenure is 5 yrs, so most PMs come in with a vision, start 2-5 programs, inherit other programs and then transition out.
The people doing the actual research are independent research labs, universities, and small business. These are a combination of scientists and engineers publishing papers and taking the technology from papers to the field.
I think you are thinking more of defense contracting than darpa.
Also, darpa has real use-cases and those use-cases tend to be hard to solve ones, they aren't VCs looking for the next trend to vampire like a national security version of the TV show shark tank.
Granted, it is 2023, it is the govt, and it is tech/eng/sci holy trinity of profit so I wouldn't be shocked if I am off the mark with my assessment.
Maybe. Even when a PM is very technical in their field, they are not expected to know the nitty gritty of the broad area they tackle in a program. To come up with realistic programs and execute them successfully, a PM needs to know where are the boundaries of current knowledge, how much $$ is required to push research to transition to real tech... and sniff out bullshit being sold by research labs. Bullshitters or not, they need to be really good at sniffing bullshit!
This is false, even more so for most Tesla models. Cars have a distributed E/E architecture where embedded software runs on several computers (called ECUs that range from anywhere between 10 and 100+). Most newer Tesla models are ahead of the curve and have a more centralized (zonal) E/E architecture with very few number of ECUs.
In summary, the infotainment system and the drivetrain definitely speak to each other and unless a secure architecture comes along, I would not discount the possibility of unintended interactions due to bugs. In fact, most famous car hacks start from the infotainment system and make their way into the drivetrain.
How is this even Zero Trust. Admittedly, there is no precise definition for ZT, but Cloudflare's solution seems to run counter to the idea of perimeter-less ZT philosophy. Instead of assuming that phones can be insecure and developing appropriate crypto based mechanisms, Cloudflare is proposing to bring the phone inside a 'trusted' network. Remember, ZT does not rely on trusted network.
Solutions like this will increase confusion and fragment the already 'interpretation led' as opposed to definition led ZT landscape.
I don't see anything about a trusted network, it looks like this is about authorizing devices. It seems a little bare on the details of how it works, but apparently it ties into a Cloudflare product called Magic WAN. Authorizing specific devices is still a good strategy even with zero-trust networking.
Device attestation is an important piece of the zero trust design, which this esim approach helps facilitate.
ZT / BeyondCorp benefits from multiple layers of security, not the hard exterior and crunchy interior approach of VPNs, and this solution from cloudflare is aligned with that.
That’s not the case — note that we don’t say “trusted network” in the blog. That’s definitely not the right solution.
There’s two key parts:
1) we can filter and secure traffic _leaving_ the device, whether bound for the Internet or internal apps. This isn’t VPN like: this is part of our software gateway. When you click (tap!) on a phishing link, we can filter it and render it inert.
2) using the eSIM, which is associated with a specific employee, as an identity signal and device posture signal. This fits squarely into the Zero Trust model. ZT is about explicit identity, not the old days of implicit “I’m on the VPN and can move laterally!”.
The takeaways are very true. I would go further to say that not only is the story telling and the scientific vision a very important part of the sell, the language and terminology needs to be in sync with the reviewers. Often times researchers write very technical details in the proposal that is hard to parse and backfires.
In non-blind proposals, pre-communication is the key.
Portfolio of a slow but precise fuzzer + fast imprecise fuzzer is the easiest integration. Start both together and return the one which fails first. SMT solvers are often complementary with samplers.
However, it would be very interesting to see if a closer integration of symbolic and sampling methodologies is possible.
