Vinnie Monaco (@vinmonaco) header
Vinnie Monaco (@vinmonaco) profile picture

Privacy engineering @OpenAI. Interested in security, privacy, artificial intelligence, and biologically-inspired computing.

SF Bay Area
Joined April 2012
💜
We have reached an agreement in principle for Sam Altman to return to OpenAI as CEO with a new initial board of Bret Taylor (Chair), Larry Summers, and Adam D'Angelo. We are collaborating to figure out the details. Thank you so much for your patience through this.
1
29
1,489
OpenAI is nothing without its people
11
22
3
434
36,235
💜
i love the openai team so much
3
9
121
18,640
Career update: I joined @OpenAI. My focus will be on data privacy engineering. I'm really looking forward to new the collaborations and challenges that lie ahead.
2
6
Vinnie Monaco retweeted
Check out our new ICML paper "Linear Adversarial Concept Erasure"! A joint work with Michael Twiton, @yoavgo and @ryandcotterell (started when visiting Ryan's awesome group in ETH Zurich!) (1/10) arxiv.org/pdf/2201.12091.pdf
2
25
9
202
Vinnie Monaco retweeted
We announce that the AISec workshop, co-located with @acm_ccs, will be held in a hybrid format (in-person + online). Deadline: June 24.  Website: aisec.cc #MLsec #MachineLearning #AI #adversarial @adversarial_ML #malware @trustworthy_ml @RedTeamVillage_
The CFP of the 15th ACM workshop on Artificial Intelligence and Security (AISec), co-located with @acm_ccs, is out! Deadline: June 24. Website: aisec.cc/ #MLsec #MachineLearning #AI #adversarial @adversarial_ML #malware @trustworthy_ml @RedTeamVillage_
A new study finds 3,000 websites on which third-party tracking companies scoop up what you type into forms in real time — even if you never hit submit. This happens on about 1,800 websites for E.U. users too, likely in violation of the GDPR. wired.com/story/leaky-forms-…
11
385
49
642
In our @IEEESP 2022 paper #NoiseSDR, @aurelsec and I build a software-defined radio transmitter out of unprivileged software and electromagnetic noise, with applications in exfiltration, tracking and injection in other receivers. s3.eurecom.fr/docs/oakland22… github.com/eurecom-s3/noise-…
1
16
1
58
Vinnie Monaco retweeted
A debate on the topic of neuromorphic computing, On the pro side: @boahen_k and Ralph Etienne-Cummings. On the not-so-pro side: Bill Dally and yours truly. March 7.
Replying to @ylecun
@ylecun will attempt to convince @boahen_k & Ralph Etienne-Cummings, where he believes the future of neural computing lies. Watch the debate live FREE on the 7th March. Save your place today: ow.ly/qVFo50I0iMa #DeepLearning #NeuralNetworks #Computation #Engineering
4
19
2
81
Proud to present this short report on the outcomes of and learnings from #NetHackChallenge21, held at @NeurIPSConf. Did DeepRL win the day, or did symbolic challengers surprise us all? What do the results tell us about next steps for AI? Read to find out! nethackchallenge.com/report.…
9
65
38
271
Vinnie Monaco retweeted
Congratulations to @yuxiwu her for #SP22 (Oakland) paper acceptance! We systematized extant human-centered work on #social #cybersecurity — i.e., work exploring S&P behaviors and threats in an ecosystem of inter-connectedness and influence. Pre-print: sauvikdas.com/papers/36/serv…
1
10
2
53
This week, I joined @Visa's security engineering team as a lead data scientist.
3
10
I was drawn by the prospect of my work having an immediate impact, continuing to focus on the problems I enjoy, and most important, an awesome team.
1
I won't be as research active in my new role but I do plan to continue some academic/gov collaboration. The team is also rapidly growing, so please DM me if you're interested in joining!!
1
Vinnie Monaco retweeted
In our new @IEEESSP 2022 paper, we measured the use of email encryption on a large longitudinal field data set (@Ch_Stransky, Oliver Wiese, Volker Roth and @yazz_acar). Our data set spans 27 years, 37k users and 81m emails.
2
21
1
59
Vinnie Monaco retweeted
ML in Cyber can have crazy constraints, like FPR < 0.001%, but how do you build your model to satisfy this? Surprisingly a lot of errors and little work studying low FP. Paper with @AndreNguyen16 @cknicholas starts to look at this problem! @IJCAIconf arxiv.org/abs/2108.04081
2
8
26
Time-based device fingerprinting demo, including some of the techniques in my upcoming @IEEESSP paper: vmonaco.com/device-fingerpri… This does incremental PSD and fundamental frequency estimation from DOM event times. Work in progress. Source/issues here: github.com/vmonaco/device-fi…
1
2
8
The browser event loop typically runs at 60Hz, trying to match screen refresh rate. But the exact Hz differs slightly between machines.
1
On the macbook air, the peak at 125Hz for mousemove events suggests the touchpad is a USB 1.1 device. There's no peak in the keydown PSD, suggesting a PS/2 connection.